US 5,751,833 AGrant
Apparatus and Method for Inspecting Articles Such as Agricultural Produce
Issue Date:1998-05-12
•6 Claims
•31 Drawing Sheets
Abstract
A complete sorting system is provided herein including a sequence of interlacing cable conveyors for conveying objects to be sorted, such as apples or other agricultural produce, from bulk storage at the input to the system, through the system, and along to drops at the output of the system. The cable conveyors are configured and arranged to handle even delicate agricultural produce without damage and at high speed. The produce or objects which are conveyed typically do not roll along the conveyors. Preferably, objects of substantially any shape can be conveyed along the system.
Metadata
Assignee
- Elop Electro-Optics Industries, Ltd.
Inventors
- Shmuel Blit
- Eyal Bartfeld
- Idan Pais
- Yair Eilam
- Efraim Vallach
- Haim Bezdin
- Israel Laron
- Doron Katzin
Application Information
Application Number:US 6548210
Filing Date:1996-05-29
Priority Date:1992-04-16
Art Unit:266
Classifications
IPC:
G06K 900
Field of Search:
382395356209348250110;155;224;28461;900;90437657689;91;129;130559.39
Patent Drawings (31 sheets)
Description
Field of the Invention
The present invention relates generally to apparatus for inspecting articles in motion and particularly to apparatus for sorting and grading agricultural produce.
Background of the Invention
In the present specification and claims, the term "sorting" is used to mean categorization of articles by size and/or color and the term "grading" is used to mean categorization of articles by quality and/or type.
Systems for sorting and grading of apples based on computing thinness ratios for blemishes are described in the following two publications by Rehkugler, G. E. and Throop, J. A.:
"Image processing algorithm for apple defect detection", Trans. of the ASAE, Vol. 32(1), pp. 267-272, 1989, and
"Apple sorting with machine vision", Trans. of the ASAE, Vol. 29(5), 1388-1397, 1986. In this reference, diffuse illumination is provided by a translucent acrylic box.
A Ph.D. dissertation by G. L. Graf ("Automatic detection of surface blemishes on apples using digital image processing", Cornell University, 1982, available from UMI Dissertation Information Service, 300 N. Zeeb Rd. Ann Arbor, Mich. 48106) discloses in FIG. 4--4 a generally cylindrical diffuser and apparatus for illuminating the diffuser from the exterior thereof, for inspecting an apple which has been manually placed within the cylindrical diffuser.
Another cylindrical diffused light system is described in Sarker, N. and R. R. Wolfe, "Computer vision based system for quality separation of fresh market tomatoes", Trans. ASAE, Vol. 28, pp. 1714-1718, 1985. The Sarker et al system includes mechanical apparatus for aligning tomatoes along the stem/calyx axis and a defect detection subsystem which employs adaptive thresholding of a gradient image.
Color inspection methods for peaches in which peaches are viewed from more than one angle using multiple sensors, a frame grabber, a CCD camera, one-sided diffuse illumination and color grading and feature extraction algorithms are mentioned in Miller, B. K. and Delwiche, M. J., "A color vision system for peach grading", Trans. of the ASAE, Vol. 32(4), 1484-1490, 1989.
An EO commercial development of a sorter and grader system employing solid-state CCD cameras in which blemishes are detected on fruit rolling on bicone rollers is described in A. Davenel et al, "Automatic detection of surface defects on fruit by using a vision system", J Agricultural and Engineering Research, 41, 1-9, 1988.
Methods for identifying bruised apple tissue using reflected light of one, two or three normalized wavelengths, including computations of differences or ratios between 2 wavelengths, are discussed in B. L. Upchurch et al, "Spectrophotometric study of bruises on whole, red delicious apples", Transactions of the ASAE, 33(2), March-April 1990.
Upchurch et al describe various models for distinguishing bruised from nonbruised areas on Red Delicious apples. One such model is set forth in Equation 5 and comprises a two wavelength difference which is divided by a normalizing wavelength to take into account illumination changes. Upchurch et al propose to cancel out differences in reflectance levels, due to factors other than tissue condition, by division.
U.S. Pat. No. 5,085,325 to Jones et al discusses color sorting of articles and use of a camera and flash to image moving articles. Color analysis is performed on the basis of RGB color imaging and spherical color coordinates.
U.S. Pat. No. 5,012,116 to Russell discusses a system for detecting defects in diffusely illuminated bearing balls which roll along a pair of parallel rails.
U.S. Pat. No. 5,010,247 to Smith et al discloses a system in which objects are dropped through a viewing zone and are viewed by three viewers arranged along 3 mutually orthogonal axes using radiation of different wavelengths.
A method for deriving information regarding convexity/concavity of an article by projecting a grid pattern on the article is disclosed in N. Shrikhande and G. Stockman, "Surface orientation from a projected grid", IEEE Transactions on pattern analysis and machine intelligence, Vol. 11(6), June 1989.
Attempts to detect fruit stems using image processing are reported in Wolf, R. R and Sander, W. E, "An algorithm for stem detection using digital image processing", Trans. ASAE, 28, pp. 641-644, 1984.
U.S. Pat. No. 4,863,041 to Bailey discloses a system for observing an article from 4 sides without imaging, using a color background which is dynamically controlled. The system is triggered as a response to intercepted light beams.
U.S. Pat. No. 5,078,258 to Van Der Shoot discloses a system which orients apples by mechanical apparatus and subsequently picks the apple. The system includes a color camera which identifies "red cheek" on an apple.
U.S. Pat. No. 4,940,850 to Satake discloses a system of 3 detectors which "look" at an article in vertical free fall through 2 spectral filters.
U.S. Pat. No. 4,878,582 to Codding discloses a bichromatic sorter in which an article is seen from 3 coplanar directions, on a color background.
U.S. Pat. No. 4,741,042 to Throop et al discloses a software system for analyzing size and shape of bruises on fruit by creating a binary image of bruises.
U.S. Pat. No. 4,645,080 to Scopatz discusses a system for grading non-orienting articles. Specifically, the system grades oranges on the basis of information from three overhead and side sensors.
U.S. Pat. No. 4,515,275 to Mills et al discloses a system for processing fruit such as lemons. Polarized illumination of the fruit is provided from 4 sides and a single video camera or line scanner is employed.
U.S. Pat. No. 4,204,950 to Burford, Jr. discloses a grading system using four spectral bands, of which two are visible.
A state of the art system for conveying fruit is marketed by Accu Pak Systems.
Published PCT Application WO 91/04803 describes apparatus for weighing, sizing and defect sorting of fruit including one or more singulators, a conveyor for passing the fruit under a CCD array camera, means for rotating the fruit under the camera, an image processor including a master processor passing signals to a vision processr and 8 object processors which divide the captured image into image sections each representing only one piece of fruit, and ejectors spaced along the conveyor which eject the fruit to outfeed conveyors corresponding to sorting categories.
Summary of the Invention
The present invention seeks to provide an improved system for inspecting articles such as but not limited to agricultural produce, including apples.
According to a preferred embodiment of the present invention, a complete sorting system is provided herein including a sequence of interlacing cable conveyors for conveying objects to be sorted, such as apples or other agricultural produce, from bulk storage at the input to the system, through the system, and along to drops at the output of the system. The cable conveyors are configured and arranged to handle even delicate agricultural produce without damage and at high speed. The produce or objects which are conveyed typically do not roll along the conveyors. Preferably, objects of substantially any shape can be conveyed along the system.
There is also provided in accordance with a preferred embodiment of the present invention apparatus for inspecting agricultural produce having a stem and a calyx including a stem/calyx identifier for determining the locations of the stem and the calyx of the agricultural produce, and a blemish detector, cooperative with the stem/calyx detector, for detecting blemishes and avoiding false detections of the stem and calyx as blemishes.
Further in accordance with a preferred embodiment of the present invention the blemish detector is also operative to ignore certain color variations in the vicinity of the stem and calyx.
Still further in accordance with a preferred embodiment of the present invention the stem/calyx identifier includes a valley contour detector.
There is additionally provided in accordance with a preferred embodiment of the present invention apparatus for inspecting and handling agricultural produce having a stem and a calyx including a stem/calyx identifier for determining the locations of the stem and the calyx of the agricultural produce, and a mechanical operator, receiving an output of the stem/calyx identifier, for carrying out a mechanical operation on the agricultural produce which takes into account the locations of the stem and the calyx.
Also in accordance with a preferred embodiment of the present invention the mechanical operator includes a cutter.
Moreover in accordance with a preferred embodiment of the present invention the mechanical operator includes a produce marker.
Further in accordance with a preferred embodiment of the present invention the mechanical operator includes a produce packer.
Still further in accordance with a preferred embodiment of the present invention the mechanical operator includes a produce sorter.
There is also provided in accordance with a preferred embodiment of the present invention apparatus for inspecting articles including a camera, and a spectral filter defining a plurality of light paths from an article to be inspected to the camera, the plurality of light paths having different spectral characteristics, wherein the spectral filter includes a prism operative to provide a plurality of images of the article having different spectral characteristics and being disposed in generally non-overlapping positions on an image plane of the camera.
There is further provided in accordance with a preferred embodiment of the present invention apparatus for inspecting articles including an enclosure formed of a generally translucent light diffusing material, an external illuminator for illuminating the enclosure from the exterior thereof, an article driver for causing articles to be inspected to pass through the enclosure, and a camera for inspecting the articles as they pass through the enclosure.
Also in accordance with a preferred embodiment of the present invention the enclosure is generally spherical.
Additionally in accordance with a preferred embodiment of the present invention the external illuminator includes a plurality of light sources surrounded by a reflector.
Further in accordance with a preferred embodiment of the present invention the enclosure and the external illuminator are operative to provide generally uniform illumination of the article as it is inspected.
Still further in accordance with a preferred embodiment of the present invention the enclosure and the external illuminator are operative to provide a generally uniform background for the article, as the article is inspected.
Also in accordance with a preferred embodiment of the present invention the article driver includes a non-contact passage provider operative to provide non-contact passage of the articles through the enclosure.
Moreover in accordance with a preferred embodiment of the present invention the external illuminator is sealed from the inside of the enclosure.
Additionally in accordance with a preferred embodiment of the present invention the camera is operative to inspect generally all exposed surfaces of the article generally simultaneously.
Further provided in accordance with a preferred embodiment of the present invention is apparatus for inspecting articles including a camera operative to inspect generally all exposed surfaces of the article generally simultaneously.
Still further in accordance with a preferred embodiment of the present invention the apparatus includes an article supporter/conveyer for supporting and conveying articles to be inspected on an article bearing element configured such that substantially insignificant portions of the surfaces of the article are obscured from inspection.
Also in accordance with a preferred embodiment of the present invention the article bearing element has a generally horizontal configuration and wherein the article supporter/conveyer also includes apparatus for driving the articles along the article bearing element.
Additionally provided in accordance with a preferred embodiment of the present invention is apparatus for inspecting articles including an IR imager for generating an IR image of an article, a color imager for generating a color image of an article, and a defect detector receiving outputs from the IR imager and from the color imager for combining the outputs and for providing an output indication of defects based on information including information derived from combining the outputs.
Further in accordance with a preferred embodiment of the present invention the article bearing element has a mesh configuration.
Still further in accordance with a preferred embodiment of the present invention the apparatus includes a camera operative to inspect generally all exposed surfaces of the article, and an image processor for providing a reconstruction of the exposed surfaces of an article in which substantially every point on the exposed surfaces of the article is reconstructed exactly once.
Also in accordance with a preferred embodiment of the present invention the apparatus include an image processor for providing a reconstruction of the exposed surfaces of an article in which substantially every point on the exposed surfaces of the article is reconstructed exactly once, wherein the camera is operative to inspect generally all exposed surfaces of an article.
There is further provided in accordance with a preferred embodiment of the present invention apparatus for inspecting articles including a camera operative to inspect generally all exposed surfaces of the article, and an image processor for providing a reconstruction of the exposed surfaces of an article in which substantially every point on the exposed surfaces of the article is reconstructed exactly once.
Further in accordance with a preferred embodiment of the present invention the camera is operative to inspect generally all exposed surfaces of an article generally simultaneously.
Still further in accordance with a preferred embodiment of the present invention the camera includes at least first and second camera units.
Also in accordance with a preferred embodiment of the present invention the image processor includes apparatus for reconstructing predetermined at least first and second substantially nonoverlapping portions of the article in accordance with the images generated by the at least first and second camera units respectively.
Also in accordance with a preferred embodiment of the present invention the predetermined portions of the article are determined in accordance with a predetermined model of the shape of the article, the model defining at least one model parameter.
Additionally in accordance with a preferred embodiment of the present invention the at least first and second camera units includes at least first, second and third camera units.
Moreover in accordance with a preferred embodiment of the present invention the camera is operative to inspect at least 70 percent of the exposed surface of an article.
Further in accordance with a preferred embodiment of the present invention the camera is operative to inspect at least 80 percent of the exposed surface of an article.
Still further in accordance with a preferred embodiment of the present invention the camera is operative to inspect at least 90 percent of the exposed surface of an article.
Also in accordance with a preferred embodiment of the present invention the camera is operative to inspect at least 95 percent of the exposed surface of an article.
Additionally in accordance with a preferred embodiment of the present invention the apparatus includes an image-model comparison unit for comparing an image of at least a portion of the article to the predetermined model of the shape of the article and for correcting the determination of the predetermined portions of the article to take into account discrepancies between the actual shape of the article and the predetermined model thereof.
Moreover in accordance with a preferred embodiment of the present invention the image processor includes apparatus for comparing first and second images of first and second respective portions of the article generated by the first and second camera units respectively, thereby to identify overlap between the first and second images.
Further in accordance with a preferred embodiment of the present invention the article is supported by an article supporting element disposed intermediate the article and the camera and wherein the image processor includes an article supporting element identifier for differentiating an image of the article supporting element from the image of the article.
Still further in accordance with a preferred embodiment of the present invention the stem/calyx identifier also includes apparatus for inspecting a putative location of the stem and of the calyx relative to the shape of the fruit and for rejecting putative locations which are not located generally one opposite the other.
There is also provided in accordance with a preferred embodiment of the present invention internal referenced apparatus for inspecting the color of an agricultural product including a spectral standard device, a product and spectral standard imager operative to image an agricultural product together with the spectral standard device, and a product-spectral standard comparator operative to determine the spectral characteristics of the agricultural product relative to the spectral standard device by comparing images thereof.
Also in accordance with a preferred embodiment of the present invention the at least one model parameter includes at least one parameter which is derived from images of individual articles.
Additionally in accordance with a preferred embodiment of the present invention the plurality of images of the article includes at least one IR image of the article.
Further in accordance with a preferred embodiment of the present invention the article bearing element includes a plurality of elongate elements in generally parallel orientation.
There is still further provided in accordance with a preferred embodiment of the present invention apparatus for inspecting agricultural produce substantially without bruising or squashing, including a camera for inspecting delicate agricultural produce, a non-contact passage provider operative to provide substantially non-contact passage of the delicate agricultural produce through the field of view of the camera, and a damper for receiving the delicate agricultural produce from the non-contact passage provider substantially without bruising or squashing the delicate agricultural produce.
Also in accordance with a preferred embodiment of the present invention the damper includes at least one brush element.
Additionally in accordance with a preferred embodiment of the present invention the damper includes an enclosure containing liquid.
Further in accordance with a preferred embodiment of the present invention the damper includes a sequencer for receiving the delicate agricultural produce in sequence and for maintaining the sequence.
Still further in accordance with a preferred embodiment of the present invention the sequencer includes a container of liquid for sequentially flushing away a sequence of agricultural products.
Also in accordance with a preferred embodiment of the present invention the article bearing element has an inclined configuration.
There is additionally provided in accordance with a preferred embodiment of the present invention apparatus for imaging articles in motion for subsequent inspection including a camera for imaging an article in motion, an article entry detector for repeatedly analyzing a small portion of the field of view of the camera in order to detect entry of the article thereto.
Moreover in accordance with a preferred embodiment of the present invention the apparatus includes a low-level illuminator for providing a low level of illumination sufficient for detecting the presence of the article, a high-level illuminator for providing a high level of illumination sufficient for inspecting characteristics of the article, and a high illumination trigger which receives an indication of the entry of the article into the small portion of the field of view from the article entry detector and triggers the high-level illuminator.
Further in accordance with a preferred embodiment of the present invention the apparatus includes an article characteristic analyzer for analyzing characteristics of the article, and an analyzer trigger for receiving an indication of the entry of the article into the small portion of the field of view from the article entry detector and for triggering the characteristics analyzer.
Still further in accordance with a preferred embodiment of the present invention the article entry detector includes apparatus for repeatedly computing a difference between a digitized first image of the field of view and a digitized subsequent image of the field of view.
There is additionally provided in accordance with a preferred embodiment of the present invention a singulation method including the steps of conveying articles in sequence without physically separating adjacently disposed articles along the conveyor.
There is further provided in accordance with a preferred embodiment of the present invention a singulation method including the steps of conveying articles in sequence without controlling the distances between adjacent article positions.
There is still further provided in accordance with a preferred embodiment of the present invention an agricultural produce inspection method including the steps of inspecting a representation of an entity forming at least a portion of an agricultural product, and employing fuzzy logic criteria in order to classify the entity according to a predetermined classification scheme.
Also in accordance with a preferred embodiment of the present invention the small portion of the field of view includes a one-dimensional array of pixels.
Additionally in accordance with a preferred embodiment of the present invention the camera is operative to inspect the agricultural produce by imaging.
Further provided in accordance with a preferred embodiment of the present invention is a method for inspecting agricultural produce having a stem and a calyx including the steps of optically determining the locations of the stem and the calyx of the agricultural produce, and detecting blemishes by cooperating with the optical determining step in order to avoid false detections of the stem and calyx as blemishes.
Still further provided in accordance with a preferred embodiment of the present invention is a method for inspecting and handling agricultural produce having a stem and a calyx including the steps of optically determining the locations of the stem and the calyx of the agricultural produce, and employing the results of the determining step in order to carry out a mechanical operation on the agricultural produce.
There is also provided in accordance with a preferred embodiment of the present invention a method for imaging articles in motion for subsequent inspection including the steps of providing a camera for imaging an article in motion, and repeatedly analyzing a small portion of the field of view of the camera in order to detect entry of the article thereto.
There is additionally provided in accordance with a preferred embodiment of the present invention a method for inspecting articles including the steps of illuminating an enclosure, formed of a generally translucent light diffusing material, from the exterior of the enclosure, causing articles to be inspected to pass through the enclosure, and inspecting the articles as they pass through the enclosure.
There is moreover provided in accordance with a preferred embodiment of the present invention a method for inspecting articles including the step of inspecting generally all exposed surfaces of the article generally simultaneously.
There is still further provided in accordance with a preferred embodiment of the present invention a method for inspecting articles including the steps of generating at least one IR image of an article, generating a color image of an article, and analyzing the images and providing an output indication of defects in the articles.
Also provided in accordance with a preferred embodiment of the present invention is an internally referenced method for inspecting the color of an agricultural product including the steps of imaging an agricultural product together with a spectral standard device, and determining the spectral characteristics of the agricultural product relative to the spectral standard device by comparing images thereof.
Additionally provided in accordance with a preferred embodiment of the present invention is a method for inspecting delicate agricultural produce substantially without bruising or squashing, including the steps of providing a camera for inspecting delicate agricultural produce, providing for substantially non-contact passage of the delicate agricultural produce through the field of view of the camera, and receiving the delicate agricultural produce following the non-contact passage substantially without bruising or squashing the delicate agricultural produce.
In accordance with a preferred embodiment of the present invention the camera includes a line scanner.
There is provided in accordance with a preferred embodiment of the present invention a method for inspecting agricultural produce including the steps of associating an indication of at least one characteristic of an individual item of agricultural produce with the individual item of produce.
Further in accordance with a preferred embodiment of the present invention the step of associating includes the step of affixing a sticker to the individual item of produce, wherein the sticker bears an indication of at least one characteristic of the individual item.
There is additionally provided in accordance with a preferred embodiment of the present invention apparatus for inspecting agricultural produce including a produce labelling unit operative to associate an indication of at least one characteristic of an individual item of agricultural produce with the item of produce, and a produce inspection unit operative to automatically inspect at least one characteristic of an individual item of produce and to provide an indication of the at least one characteristic to the produce labelling unit.
Further in accordance with a preferred embodiment of the present invention the at least one characteristic includes at least one of the following characteristics variety, size, weight, grade, color, and price.
There is also provided in accordance with a preferred embodiment of the present invention a singulator including a conveyor configured and arranged to convey articles in sequence, characterized in that no element is provided to separate adjacently disposed articles along the conveyor.
There is additionally provided in accordance with a preferred embodiment of the present invention a singulator including a conveyor configured and arranged to convey articles in sequence without controlling the distances between adjacent article positions.
Further in accordance with a preferred embodiment of the present invention the conveyor includes a plurality of elongate elements arranged in the direction in which the articles are conveyed.
Still further in accordance with a preferred embodiment of the present invention the plurality of elongate elements includes a plurality of cables.
There is further provided in accordance with a preferred embodiment of the present invention a singulator including a conveyor configured and arranged to convey articles in sequence, wherein the conveyor includes a plurality of strings of beads.
There is also provided in accordance with a preferred embodiment of the present invention a method for inspecting agricultural produce including the steps of generating an image of individual agricultural products, and automatically grading agricultural products by receiving the image generated by the imager and providing an output indication of a grade for the individual agricultural product.
There is further provided in accordance with a preferred embodiment of the present invention an agricultural produce inspection system including a fuzzy logic image entity classifier which employs fuzzy logic criteria in order to inspect a representation of an entity forming at least a portion of an agricultural product and to classify the entity according to a predetermined classification scheme.
Additionally in accordance with a preferred embodiment of the present invention the agricultural product is expected to have at least one discolored area on its surface and wherein the entity includes a discolored portion of the agricultural product and wherein the predetermined classification scheme includes two classes: expected discoloration, and unexpected discoloration.
Further in accordance with a preferred embodiment of the present invention the entity includes a pair of discolored areas and the predetermined classification scheme includes a binary classification scheme including a first class of stem/calyx and a second class of not stem/not calyx.
Still further in accordance with a preferred embodiment of the present invention the predetermined classification scheme includes a plurality of classes corresponding to a plurality of types of blemishes.
Also in accordance with a preferred embodiment of the present invention the entity includes the agricultural product and wherein the predetermined classification includes a plurality of classes corresponding to a plurality of grades of agricultural produce.
Additionally in accordance with a preferred embodiment of the present invention the representation of the entity is derived by employing fuzzy logic criteria in order to inspect a representation of at least one subentity forming a portion of the entity.
There is also provided in accordance with a preferred embodiment of the present invention an agricultural produce inspection system including an imager for generating an image of individual agricultural products, and an automatic agricultural product grading unit including an image processor receiving the image generated by the imager and providing an output indication of a grade for the individual agricultural product.
Further in accordance with a preferred embodiment of the present invention the grading unit is user-tunable such that criteria employed by the grading unit may be modified by the user.
Still further in accordance with a preferred embodiment of the present invention the grading unit includes a user-tunable fuzzy logic unit.
Also in accordance with a preferred embodiment of the present invention the fuzzy logic image entity classifier includes an entity representation fuzzification unit, entity classification rule inference logic, and an entity classification defuzzification unit.
Additionally in accordance with a preferred embodiment of the present invention a plurality of classification rules are employed, each rule corresponding to an individual class and having associated therewith, for each entity to be classified, a degree of belief, wherein the entity classification defuzzification unit is operative to identify, for each entity, a classification rule having a maximal degree of belief and to associate the entity with the class corresponding to the classification rule having the maximal degree of belief.
There is further provided in accordance with a preferred embodiment of the present invention apparatus for detecting blemishes on agricultural produce, the apparatus including a dual IR illumination system operative to provide IR band I and IR band II illumination of the agricultural produce, and a sensor operative to receive first and second images of the agricultural produce as illuminated by the IR band I and IR band II illumination respectively, a comparison operator operative to compare the first and second images and to generate an output indication of blemishes, wherein the wavelength bands of the band I IR and band II IR illuminations are selected such that artifacts resembling blemishes are similarly detected by the two illuminations whereas blemishes are differently detected in the two illuminations.
Still further in accordance with a preferred embodiment of the present invention the wavelength band of the band I IR illumination includes approximately 0.67-0.77 microns and the wavelength band of the band II IR illumination includes approximately 0.77-0.95 microns.
Moreover in accordance with a preferred embodiment of the present invention the comparison operator includes an arithmetic operation.
Further in accordance with a preferred embodiment of the present invention the article includes an agricultural produce.
Still further in accordance with a preferred embodiment of the present invention the agricultural produce includes an apple.
Also in accordance with a preferred embodiment of the present invention the article includes a round article.
Additionally in accordance with a preferred embodiment of the present invention the agricultural produce includes a round article.
There is also provided in accordance with a preferred embodiment of the present invention conveying apparatus for agricultural produce including a produce supporting cable assembly including a plurality of cables arranged to support agricultural produce.
Further in accordance with a preferred embodiment of the present invention the conveying apparatus includes an additional cable assembly interlaced with the produce supporting cable assembly.
Still further in accordance with a preferred embodiment of the present invention the conveying apparatus includes a kicker element, and a kicker element activator operative to slide the kicker element between the produce supported by the cable assembly so as to allow the kicker element to engage the produce and to remove the produce from the plurality of cables.
Also in accordance with a preferred embodiment of the present invention the spectral filter includes a non-pyramidal prism.
Brief Description of the Drawings
The present invention will be understood and appreciated from the following detailed description, taken in conjunction with the drawings in which:
FIG. 1 is a block diagram of an electro-optic sorting and grading system constructed and operative in accordance with a preferred embodiment of the present invention;
FIG. 2 is a block diagram of a sorting and grading line constructed and operative in accordance with a preferred embodiment of the present invention and including the electro-optic sorting and grading system of FIG. 1;
FIG. 3A is a side view illustration of conveying apparatus for conveying or transporting agricultural produce substantially without contacting or obscuring the agricultural produce during imaging, useful in the embodiment of FIG. 1, which is constructed and operative in accordance with a first preferred embodiment of the present invention;
FIG. 3B is a cross sectional schematic illustration of an article receiving system which is a portion of the apparatus of FIG. 3A and is constructed and operative in accordance with a second preferred embodiment of the present invention;
FIG. 4A is a schematic illustration of conveying apparatus useful in the embodiment of FIG. 1 which is constructed and operative in accordance with a preferred embodiment of the present invention wherein agricultural produce rolls down an inclined plane defined by a plurality of elongate elements having a downward projection of relatively small area;
FIGS. 4B and 4C are cross-sectional illustrations of the plurality of elongate elements of FIG. 4A, arranged in accordance with alternative embodiments of the present invention;
FIG. 5 is a perspective illustration of a portion of the apparatus of FIG. 4A;
FIG. 6 is a side view illustration of conveying apparatus similar to the apparatus of FIG. 4A but wherein the plurality of generally parallel elogate elements is replaced by an element having a screen or mesh type configuration;
FIG. 7 is a schematic illustration of conveying apparatus useful in the embodiment of FIG. 1 which is constructed and operative in accordance with a preferred embodiment of the present invention wherein agricultural produce to be imaged is actively conveyed along a generally horizontal pathway defined by a plurality of elongate elements having a downward projection of relatively small area;
FIG. 8 is a perspective illustration of apparatus similar to the apparatus of FIG. 7 except that 4 pulleys are employed rather than 3 pulleys as in FIG. 7;
FIG. 9 is a side view illustration of conveying apparatus similar to the apparatus of FIG. 7 but having a screen or mesh type configuration rather than an elongate configuration;
FIG. 10 is a perspective view partially cut away illustration of a diffuse illumination and imaging enclosure useful in the embodiment of FIG. 1;
FIG. 11 is an optical diagram of the illumination apparatus of FIG. 10;
FIG. 12 is a schematic illustration of the fields of view of each of the cameras of the imaging apparatus of FIG. 10;
FIG. 13 is a perspective view partially cut away illustration of a generally truncated spherical diffuse illumination and imaging apparatus useful in the embodiment of FIG. 1;
FIG. 14 is a cross sectional illustration of the apparatus of FIG. 13;
FIG. 15 is a pictorial illustration of an article in motion within the illumination and imaging apparatus of FIGS. 10-12 and a method for triggering illumination or article inspection when the article in motion reaches a predetermined position along its trajectory;
FIG. 16 is a pictorial illustration of a spectral standard useful in conjunction with the illumination and imaging apparatus of FIGS. 10-14;
FIG. 17 is an optical schematic diagram of an electro-optical imaging unit forming part of the apparatus of FIGS. 10 and 13, constructed and operative in accordance with a preferred embodiment of the present invention;
FIG. 18 is a perspective illustration of a spectral splitting element which is useful in conjunction with the apparatus of FIG. 17, which is constructed and operative in accordance with a first preferred embodiment of the present invention;
FIG. 19 is a cross sectional illustration of the apparatus of FIG. 18;
FIG. 20 is a perspective illustration of a spectral splitting element which is useful in conjunction with the apparatus of FIG. 17, which is constructed and operative in accordance with a second preferred embodiment of the present invention;
FIG. 21 is a pictorial illustration of an image generated by imaging an apple and a spectral standard using the apparatus of FIG. 18;
FIG. 22 is an optical schematic diagram of a light projector forming part of the apparatus of FIGS. 10 and 13;
FIG. 23 is a simplified flowchart of a preferred method for implementing the image processing unit 18 of FIG. 1;
FIG. 24 is a simplified flowchart of a preferred method for finding the boundary of an article using an image of the article and a grid projected onto the article which is suitable for implementing step 1816 of FIG. 23;
FIG. 25 is a simplified flowchart of a preferred method for implementing step 1820 of FIG. 23;
FIG. 26 is a simplified flowchart of a preferred method for generating apple variety data to be loaded in step 1810 of FIG. 23;
FIG. 27 is a pictorial illustration illustrating a preferred method for implementing step 1828 of FIG. 23;
FIG. 28 is a graph of spectral characteristics of each of a plurality of filters employed in the apparatus of FIG. 18;
FIG. 29 is a perspective illustration of the conveying apparatus of FIG. 7 in operative association with the imaging apparatus of FIG. 10;
FIG. 30A is a simplified side view illustration of singulating apparatus including a conveying junction comprising an interlaced pair of conveyors;
FIG. 30B is a detailed cross-sectional illustration of a preferred implementation of one of the roller assemblies of FIG. 30A;
FIG. 31A-31C illustrate apparatus for selectably ejecting articles from a cable conveyor onto a selected one of a plurality of bins corresponding to a plurality of categories into which the apples have been categorized;
FIG. 32 illustrates a preferred alternative to the conveying apparatus of FIG. 7;
FIG. 33A and FIG. 33B illustrate interlacing cables 2240 and 148 of FIG. 32; and
FIGS. 34A and 34B, taken together, form an electronic schematic diagram of components which are not commercially available and which are useful in conjunction with the Appendices attached herewith.
Also submitted herewith are the following appendices:
Appendix A is a computer listing of a software implementation of a preferred method for implementing steps 1813 and 1814 of FIG. 23;
Appendix B includes a computer listing of a main program for an article inspecting system constructed and operative in accordance with a preferred embodiment of the present invention;
Appendix C is a computer listing of a software implementation of a sample method for implementing the method of FIG. 24;
Appendix D is a computer listing of a software implementation of a sample method for assigning identifying color codes to each of a plurality of locations in imaged StarKing apples;
Appendix E is a computer listing of a software implementation of a sample method for implementing the method of FIG. 25;
Appendix F is a computer listing of a software implementation of a preferred method for implementing steps 1824 and 1828 of FIG. 23;
Appendix G is a computer listing of a software implementation of a preferred method for implementing steps 1830 and 1832 of FIG. 23;
Appendix H, taken together with Appendix I is a computer listing of a software implementation of a preferred method for implementing step 1834 of FIG. 23;
Appendix I includes a computer listing which, taken together with Appendix H, forms a computer listing of a software implementation of a preferred method for implementing step 1834 of FIG. 23, and also includes a computer listing of a software implementation of a preferred method for implementing steps 1836 and 1838 of FIG. 23.
Appendix J includes computer listings of helper routines useful in conjunction with the other computer listings appended hereto;
Appendix K includes computer listings of "include files" useful in conjunction with the other computer listings appended hereto; and
Appendix L is a computer listing of a batch file useful in generating executable programs from the other computer listings appended hereto, and a DOS configuration file.
Detailed Description of Preferred Embodiments
Reference is now made to FIG. 1 which is a block diagram of an electro-optic sorting and grading system referenced generally 10 which is constructed and operative in accordance with a preferred embodiment of the present invention. The apparatus of FIG. 1 comprises a conveying unit 12 which receives a sequence of articles to be sorted and graded and conveys the articles past an electro-optical imaging system 14. Preferably, conveying unit 12 conveys the articles so as to allow simultaneous imaging of substantially every point along the surfaces of the articles. Conveying unit 12 may receive singulated articles or may receive articles in bulk. Various embodiments for conveying unit 12 are described below with reference to FIGS. 3A-9 and 30-33.
Electro-optical imaging system 14 preferably includes a plurality of simultaneously operative electro-optical imaging units for imaging each article in the sequence of articles from a corresponding plurality of angles. Preferably, each imaging unit comprises a monochromatic camera unit and a spectral splitting or spectral separation element. Each monochromatic camera unit may comprise any suitable camera apparatus such as one or more line scanner cameras or such as, in the illustrated embodiment, as explained below with reference to FIG. 17, a CCD, CID or vidicon camera. The spectral splitting or spectral separation element, typically a prism, is operative to provide a plurality of non-overlapping images of each article, each image having different spectral characteristics. Imaging system 14 also preferably includes an illumination subsystem including a plurality of light sources, provided externally of an article-containing enclosure and sealed from the inside thereof.
The light sources are preferably surrounded by a diffusive reflector which provides generally uniform illumination of the article and of the internal surface of the article-containing enclosure which serves as a background when the article is imaged. Preferably, the article-containing enclosure is formed of a translucent material which diffuses light arriving from the light sources, thereby enhancing uniformity of illumination of the article. Electro-optical imaging system 14 is described in detail below with reference to the embodiments of FIGS. 10 and 13.
The images of each article generated by imaging system 14 are provided to an image processing unit 18 which is described below with reference to FIG. 23. Image processing unit 18 is operative to combine the outputs of the plurality of imaging units in order to quantify a plurality of features of the inspected articles. Preferably, almost every point along the surface of each article is represented exactly once in the analysis of image processing unit 18.
The feature information generated by image processing unit 18 is received by an article data analysis and sorting/grading/labelling unit 20. Analysis and sorting/grading/labelling unit 20 comprises a sorting and grading decision tree and is operative to categorize articles using a plurality of predetermined categories which may be based on size and/or shape and/or spectral characteristics such as color and/or presence of blemishes upon the articles.
Preferably, unit 20 is also operative to label or otherwise mark each individual article with an indication of an individual feature thereof as determined by the sorting and grading decision tree, such as the grade thereof.
It is appreciated that the apparatus of FIG. 1 may be used to form part of a fruit sorting and grading line such as the fruit sorting and grading line 50 of FIG. 2. As shown, fruit sorting/grading line 50 may include some or all of the following units in addition to the sorting and grading system 10 of FIG. 1, not necessarily in the illustrated sequence: an unloading unit 52, a fruit washing unit 54, a fruit drying tunnel 56, a fruit feeding unit or bulk conveyor 58, a singulator 60, optionally, and conveyor belts 62 for conveying the fruit to outlets for sorted, graded and, preferably, labelled fruits.
Singulator 60 is operative to sequence a stream of fruit by creating parallel "lanes" 64 of fruit. Each fruit preferably has a well-defined position such as between two particular bicone rollers or within a particular cup or pocket. Typically, one electro-optical imaging unit 14 is provided per lane and the configuration of electro-optical imaging unit 14 is such that the lanes may be disposed relatively close together, as explained below in more detail with reference to FIG. 13.
A mechanical kicker or air jet or other suitable means may be provided (not shown) which, in response to a suitable command from electro-optical unit 14, kicks or otherwise conveys an apple to an individual one of outlet conveyors 62.
FIG. 3A is a side view illustration of conveying unit 12 of FIG. 1 which is constructed and operative in accordance with a first preferred embodiment of the present invention. A particular feature of the embodiment of FIG. 3A is that the articles to be inspected are conveyed substantially without contact therewith during imaging, thereby minimizing obscuring of the articles to be inspected. As shown, the apparatus of FIG. 3A comprises an imaging space 80 in which the articles move substantially without being contacted and therefore substantially without being obscured from imaging apparatus. The imaging space 80 may be enclosed by an enclosure 81 which may be identical to enclosure 230 of FIG. 10 or enclosure 300 of FIG. 13, both of which are described in detail below with reference to FIGS. 10 and 13 respectively.
A conveying and launching unit 82, typically configured as an endless driving chain, conveys articles, such as fruit arriving from lane conveyor 64 of FIG. 2, toward the imaging space 80 and provides horizontal velocity to the articles, thereby enabling them to move in a generally parabolic trajectory through the imaging space 80. An unloading unit 84 receives articles which have travelled through imaging space 80 substantially without damaging the articles upon impact. Unloading unit 84 may also convey the articles to a desired location, such as to conveyor 62 of FIG. 2. Conveying and launching unit 82 may comprise an array of article supporting elements 86 such as but not limited to cylindrical or biconical rollers, rubber cups, plastic cups or trays, or flexible pockets.
Unloading unit 84 preferably comprises a damping element 96, for receiving a sequence of articles following passage of the articles through imaging space 80, and a conveyor 97. Damping element 96 typically receives the articles without substantially damaging the article upon impact, and conveyor 97 then conveys articles onward while maintaining the sequence in which they were received. Conveyor 97 may be similar to conveying and launching unit 82.
According to one preferred embodiment of the present invention, damping element 96 may comprise a brush array 98. Brush array 98 may include two or more cylindrical brushes arranged parallel to one another and rotating in opposite directions at a suitable speed such as 45 rpm. Any suitable size and type of brush may be employed, such as a 25 cm diameter, 20 cm long nylon brush with split hair so as not to damage delicate produce. The distance between axes of adjacent cylindrical brushes may be approximately 50 cm.
Alternatively, damping element 96 may comprise deformable cushions or bean bags (not shown) for damping articles, such as those described in U.S. Pat. No. 3,961,701 to Paddock et al, the disclosure of which is incorporated herein by reference.
Reference is now made to FIG. 3B which is a cross-sectional illustration of a portion of conveying unit 12 which is operative to convey agricultural produce away from electro-optical imaging system 14. The embodiment of FIG. 3B is substantially identical to the embodiment of FIG. 3A except that brush array 98 is replaced by a liquid based damping unit 100 and unloading unit 84 is replaced with a liquid-immersed unloading unit 102 which may be substantially similar to unloading unit 84 except as specified hereinbelow:
Liquid based damping unit 100 preferably comprises an open enclosure portion 103 within which airborne articles land one at a time in a liquid such as water.
Unloading unit 102 is immersed in approximately 30 cm of water and includes cups 101, or any other suitable element for scooping up articles from the water, which are configured so as to prevent disengagement of the articles from the cups due to water resistance to motion of the articles and of unloading unit 102. Preferably, cups 101 are perforated so that water may drain as articles are conveyed out of the water.
An array of liquid jet outlets 104 is arranged along the interior walls of enclosure 103 in fluid communication with a liquid circulator 106 via a piping system 107. Liquid circulator 106 may comprise a high capacity centrifugal pump which generates liquid circulation at a suitable rate such as 5 liter/sec.
Liquid outlets 104 provide liquid fountains or streams which are operative to flush away an article which has landed within enclosure 103 before the next article lands.
Articles are flushed along a predetermined trajectory, in sequence, through a conduit portion 108 toward unloading unit 102 and eventually are conveyed out of damping unit 100 in sequence by unloading unit 102. The operation of unloading unit 102 is typically synchronized relative to the flow of articles to the system, so as to scoop up each article as it flows toward unloading unit 102. Preferably, the velocity of the article relative to the velocity of unloading unit 102 at the moment the article is scooped up by unloading unit 102 is zero. For apples and other articles which are lighter than water, the article has an upward component which is matched by the upward velocity component of unloading unit 102.
Reference is now made to FIGS. 4A-4C and 5 which illustrate conveying unit 12 of FIG. 1 constructed and operative in accordance with a third preferred embodiment of the present invention. In the embodiment of FIGS. 4A-4C and 5, agricultural produce rolls down an inclined pathway defined by a plurality of elongate elements which are configured to conceal a very small portion of the agricultural produce from imaging system 14 of FIG. 1.
As shown, the apparatus of FIGS. 4A-4C and 5 comprises an inclined unit 110, a loading unit or singulator 112 and an unloading unit 114 which may be similar to unloading unit 84 of FIG. 3A. Singulator 112 may comprise a roller conveyor as shown or a cups conveyor such as conveyor 82 in FIG. 3A or more generally any suitable conveying element such as a conveyor forming part of an existing sorting and grading line.
Inclined unit 110 typically includes a plurality of cables 120 which are arranged substantially parallel to one another and to the path along which the articles are to be conveyed. Any suitable number of suitable spaced cables may be employed, such as three cables spaced 20-40 mm apart. Each cable may be formed from any suitable material which is preferably relatively strong so as to allow the cross sectional diameter of the cables to be small, such as approximately 1-3 mm, thereby minimally obscuring the articles when viewed through the cables. For example, cables may be formed of nylon or plastic coated steel, or of plastic coated polysterene or aramid fibers. The length of cables 120 may be approximately 100-120 cm.
As shown in FIGS. 4B and 4C, the cables 120 when viewed in cross section may for example be arranged at equal intervals along an approximately 120 degree arc of an 8 cm diameter circle.
Cables 120 are typically arranged at a suitable angle such as approximately 20-40 degrees from the horizontal. The cables may be supported by a rigid frame 122 which maintains a suitable level of tension such as approximately 50 kg by means of suitable tensioning elements such as an array 123 of springs. The articles, as they roll down the cables, gain velocity and the angle and length of the cables are preferably selected such that only one article is in the field of the view of the imaging cameras at any one time.
Loading unit 112 is positioned as close as possible to the high end 124 of the cables, at an elevation the same as or slightly exceeding the height of the high end 124 of the cables, so that articles will tend to be deposited from loading unit 112 directly onto the cables 120.
FIG. 6 illustrates an alternative embodiment of conveying unit 12 of FIG. 1. The embodiment of FIG. 6 is similar to the embodiment of FIGS. 4 and 5, however, the cables 120 are replaced by a net 130 arranged at a suitable angle to the horizontal, such as 40-50 degrees. The articles roll down the length of the net, which may be approximately 110-130 cm long. Loading onto the net and unloading therefrom may be as described above with reference to FIGS. 4A-4C and 5.
Any suitable net or mesh may be employed which obscures the articles as little as possible when the articles are viewed through the net. For example, a net or mesh may be formed of nylon threads 0.3-0.8 mm in diameter and spaced appproximately 20 to 30 mm apart. The net 130 may be supported by a rigid frame 132 which preferably includes means, such as an array 136 of springs, for maintaining the net 130 at a level of tension sufficient to ensure that articles rolling along the net depress the net by no more than approximately 2 cm.
According to one embodiment of the present invention, the frame 132 comprises a pair of net-supporting cables 134 disposed on both sides of the net 130. Cables 134 may be approximately 20 cm apart and provide a suitable level of tension such as approximately 20 kg such that, when the net 130 is supporting an article, the lowest point of the net 130 is approximately 3 cm below the cables 134.
Alternatively, the cables may be eliminated and tension in the net may be provided by springs 136.
FIGS. 7 and 8 illustrate an alternative embodiment of conveying unit 12 of FIG. 1. The embodiment of FIGS. 7 and 8 is similar to the embodiment of FIGS. 4A-C and 5 except for the following differences. The cables 120 of FIGS. 4A-C and 5 are replaced by driven cables 148 which may be oriented generally horizontally rather than being arranged along an inclined plane.
Typically, the angle between cables 148 and the horizontal is within the range of +/-5 degrees. Cables 148 are preferably closed loops formed of a suitable material such as plasticized PVC or such as steel or polyester or aramid fibers coated with plasticized PVC or nylon or polyurethane.
A cable supporting unit 150 is provided for tensioning the pulleys at about 30-50 kg. Cable supporting unit 150 preferably comprises an array of coaxial guiding pulleys 152 comprising outer pulleys 154 and inner pulleys 156. Inner pulleys 156, which support central cables 155, are typically smaller than outer pulleys 154, which support peripheral cables 157, such that central cables 155 are supported below peripheral cables 157, providing a cradle-like configuration of cables 148. For example, outer pulleys 154 may have diameters of approximately 100 mm and inner pulleys 156 may have diameters of approximately 70 mm. Each pulley is preferably independent or disengaged relative to the other pulleys.
A cable driving unit 160 is provided for driving the cables 148. Cable driving unit 160 preferably comprises an array 162 of drive/power pulleys which are coaxially mounted on an electric motor shaft 164. Preferably, all drive/power pulleys have the same diameter, such as approximately 100 mm, in order to maintain a uniform linear speed for all of cables 148, such as approximately 1.2-1.4 m/sec.
Preferably, a loading unit or singulator 170 is provided which is generally similar to singulator 112 of FIGS. 4A-5. However, loading unit 170 is configured and arranged relative to cables 148 such that, once loaded onto the cables, articles remain generally stationary relative to the cables and do not roll. Also, preferably, the cables move the articles such that only one article at a time is within the field of view of the imaging apparatus. The velocity of loading unit 170 is preferably approximately 50 cm/sec.
An unloading unit 172 is provided which may be similar to unloading unit 114 of FIG. 4A.
FIG. 9 illustrates an embodiment of conveying unit 12 of FIG. 1 which is a variation on the embodiment of FIGS. 7 and 8. The embodiment of FIG. 9 is similar to the embodiment of FIGS. 7 and 8 except for the following differences.
The cables 148 are replaced with a net 180 which may be similar to net 130 of FIG. 6 and which may be supported at a tension level of approximately 10 kg.
According to one embodiment of the present invention, cables 190 which are similar to cables 134 of FIG. 6 support both sides of the net and are supported by pulleys 192 associated with a motor 194. Motor 194 via pulleys 192 convey the cables 190 in endless motion at a suitable linear velocity such as approximately 0.4 m/sec.
According to an alternative embodiment of the present invention, cables 190 are omitted. In this embodiment, the tension of the net 180 is relatively low, such as approximately 5 kg.
The linear speed at which articles are conveyed along the net may be approximately 0.4 m/sec.
A particular feature of the embodiments of FIGS. 3A-9 is that a large proportion of the surface area of the articles to be inspected is unobscured from any angle and may be imaged, such as 80%, 90%, 95% or more of each article. A plurality of imaging units may be employed such as two, three, four or more imaging units arranged to image the article from different angles.
Reference is now made to FIG. 10 which is a perspective view partially cut away illustration of electro-optical imaging system 14 of FIG. 1 constructed and operative in accordance with a preferred embodiment of the present invention. The apparatus of FIG. 10 preferably includes a plurality of simultaneously operative electro-optical imaging units for imaging each article in the sequence of articles from a corresponding plurality of angles. In the illustrated embodiment, four electro-optical imaging units are provided of which three are shown, referenced 210, 212 and 214. Preferably, each imaging unit includes a camera unit 216 described in detail below with reference to FIG. 17.
Optionally, associated with each imaging unit is a light projector 218 equipped with a grid-shaped mask (not shown) which projects a grid image onto the article and onto a background surface such as the inner surface of an imaging enclosure 230, so that the camera unit 216 may image the grid image superimposed onto each article. The lines or stripes of the grid, when superimposed on the article, are distorted as a function of convexity/concavity of the article. Therefore, by projecting a predetermined pattern, such as a grid pattern, onto an article, convexity/concavity information regarding the article may be derived. Specifically, grid projection is useful in detecting local depressions or "valleys" on an otherwise generally spherical surface, which valleys are putative locations of a fruit stem or calyx.
Methods for analyzing 3D surfaces using projected light grids are discussed in the following reference, the disclosure of which is incorporated herein by reference: Hu, G. and Stockman, G. "3D surface solution using structured light and constraint propagation", IEEE Trans. on PAMI, Vol. 4, 390-402, 1989.
A typical light projector is described in detail below with reference to FIG. 22.
Articles may travel into enclosure 230 for imaging through an entrance opening 234 and may exit enclosure 230 after being imaged through an exit opening 236. Preferably a pair of imaging apertures 237 and 238 (FIG. 12) each of suitable shape and size such as a 7 cm diameter circle are defined in enclosure 230 for each imaging unit. Apertures 237 and 238 provide optical communication between the interior of imaging enclosure 230, and camera unit 216 and optional grid projector 218, respectively, of each imaging unit. Each imaging aperture may be fitted with an optical glass window 239 (FIG. 12) of suitable thickness such as 3 mm. Each window 239 may be attached, preferably sealingly attached, to the enclosure 230 by any suitable means. For example, windows 239 may be glued into apertures 237 and 238 or alternatively may be attached by means of O-rings and brackets.
The apparatus of FIG. 10 also preferably includes an illumination subsystem including a plurality of light sources 240, provided externally of generally spherical enclosure 230 and sealed from the inside thereof. Any suitable number of light sources may be provided, such as 18 50 W light sources, generally symmetrically arranged along the internal surface of a diffusive reflective housing 250 so as to be generally equidistant from the surface of enclosure 230 and so as to be as distant as possible therefrom, thereby enhancing diffuseness of illumination and providing generally uniform illumination of the article. The light sources may comprise conventional incandescent, tungsten-halogen, fluorescent, rare-earth xenon or krytpon arc lamps or flash lamps and are preferably operative to provide a relatively flat spectrum of light ranging from green to near infra-red.
The imaging units 210, 212 and 214 are preferably supported by the housing 250 or alternatively, as shown, may be supported by enclosure 230.
The diffusive reflector housing 250 may be formed of structural panels of metal, plastic or wood which are coated from the inside with a white scatterer/ diffuser material such as white alkyd non-glossy paint or white laminated thermosetting panels. Typically, cooling fans or other cooling means (not shown) are provided externally to housing 250 to cool the sealed space between enclosure 230 and housing 250. Alternatively or in addition, fanning or ventilation of the sealed space are provided.
According to a preferred embodiment of the invention, enclosure 230 is formed of a translucent material such as ground glass, opal glass or a plastic such as polycarbonate, polyethylane, high impact polystyren or acrylic. Enclosure 230 diffuses light arriving from light sources 240, thereby enhancing uniformity of illumination of articles inside enclosure 230.
The diffusive reflector housing 250 also preferably provides generally uniform illumination of the internal surface of the enclosure 230 thereby providing a generally uniformly illuminated background for imaging of articles inside the enclosure 230.
A suitable diameter for enclosure 230 is approximately 60 cm. Suitable dimensions for housing 250 is 80 cm.times.80 cm.times.80 cm. It is appreciated that the housing 250 need not be in the shape of a cube. Any suitable shape may be employed, such as a sphere or right cylinder having circular or polygonal cross-section. According to a preferred embodiment of the present invention, the cross section of the right cylinder, which defines side surfaces of the imaging enclosure, is a circle or a polygon having a number of sides which is an integer multiple of the number of imaging units, for reasons of symmetry. For example, if three imaging units are employed, the cross-section may be an equilateral hexagon.
FIG. 11 is an optical diagram of the illumination apparatus of FIG. 10. As shown, illumination is preferably provided from a plurality of symmetric directions such as four symmetric directions, if four coplanar imaging units are employed.
FIG. 12 is a schematic illustration of the intersections 270 of the internal surface of enclosure 230 with the fields of view of the four imaging units of FIG. 10. Each of the four intersections 270 serves as a background on which the article is imaged by the corresponding imaging unit. The backgrounds are preferably back-illuminated by light travelling between spherical enclosure 230 and housing 250, and further illuminated by light scattered within enclosure 230, as explained above. Preferably, the backgrounds are homogeneously bright within a range of +/-5%.
Preferably, each background is brighter than the brightest location on the article because the backgrounds, and not the articles, are backlit. The difference in lighting between article and background facilitates easy distinction between article and background in image processing.
Optionally, a spectral standard, serving as an internal, on-line imaged reference to which the image of the article may be compared, is placed along the interior surface of enclosure 230, within each of backgrounds 270. Each spectral standard is positioned at a suitable location such as adjacent the edge of the respective background 270 such that the corresponding imaging unit may simultaneously image an article to be inspected and the spectral standard. Each spectral standard may be reflective or transparent and may be attached as by an adhesive to the internal surface of the imaging enclosure 230. A sample spectral standard is described below with reference to FIG. 16.
FIGS. 13 and 14 are a perspective view partially cut away illustration and a cross sectional illustration, respectively, of apparatus which is a variation of the apparatus of FIG. 10. The apparatus of FIG. 13 is similar to the apparatus of FIG. 10, however, spherical enclosure 230 is replaced by an enclosure 300 which resembles spherical enclosure 230 except that its configuration is that of a truncated sphere. The circumference 302 of the circle of truncation may be approximately 65 percent of the circumference 310 of the spherical portion of truncated sphere 300. Imaging units may be positioned at equal intervals along the circumference 310.
Preferably, the enclosure 300 has a plane of symmetry defined by circumference 310 on which three imaging units are centered. In this embodiment, the two locations intersecting an axis 312 which is perpendicular to the plane defined by circumference 310 are typically not imaged and therefore, relatively poor illumination of these locations does not adversely affect operation.
According to a preferred embodiment of the present invention, images of the articles travelling through enclosure 230 of FIG. 10 or enclosure 300 of FIG. 13 which are employed for subsequent analysis of the articles are generated in accordance with a predetermined time schedule. The predetermined time schedule ensures that at the moment of imaging, the article being imaged is correctly positioned relative to the imaging units.
FIG. 15 is a pictorial illustration of an article in motion within the illumination and imaging apparatus of FIGS. 10-13 and a method for triggering the imaging units when the article in motion reaches a predetermined position along its trajectory. The method of FIG. 15 may be employed for triggering a high level of illumination or for triggering the imaging-for-analysis of the article in motion or for triggering any other system response.
Optionally, one or more imaging unit digitizes a predetermined portion of its field of view generally continuously, whereas the remainder of the imaging units only digitize their fields of view upon triggering by the continuously digitizing imaging unit. The continuously digitzing imaging unit may trigger the remainder of the imaging units either directly or via a central control mechanism such as a computer. The continuously digitizing imaging unit preferably digitizes at a relatively low resolution when digitizing in order to monitor article position and digitizes at a relatively high resolution when digitizing for analysis.
When digitizing in order to monitor article position, the imaging unit preferably analyzes only a single column or row of the field of view, such as a column close to the left boundary of the field of view, if the article enters the field of view from the left, or a row close to the top boundary of the field of view, if the article enters the field of view from the top. Preferably, a row or column which is near the relevant boundary but not so near as to be subject to edge effects, if present, is selected, such as the fourth row or column. Triggering occurs if a predetermined result is obtained from digitization of the monitored row or column in an individual frame, such as presence of at least ten consecutive pixels having reflectance values which differ from the reflectance values of the same pixels in the previous frame.
FIG. 15 shows the trajectory 330 of an object, which may have any shape such as a substantially straight line, in relation to the field of view 332 of an imaging unit (not shown), and three positions 340, 342 and 344 of the object along the trajectory 330, which the object reaches at times t.sub.1, t.sub.2 and t.sub.3, respectively. At time t.sub.1, the object is external to the field of view 332 and therefore, monitoring of the fourth column from the left, referenced 346, of the field of view, will not result in triggering. At time t.sub.2, the object has reached a vertical location corresponding to the fourth column 346 of the field of view 332 and therefore, triggering will result in the time interval between t.sub.2 and t.sub.3. At time t.sub.3, the object has reached the center 344 of the field of view of the imaging unit and is imaged for analysis by all imaging units (not shown).
A particular feature of the method of FIG. 15 is that only a small portion of the field of view of the illumination and imaging apparatus, and only a small portion of the trajectory of the articles in motion, need be monitored. Also, only a single one of a plurality of imaging-for-analysis units need be employed in implementing the method of FIG. 15. The individual unit may trigger operation of the remaining units when the article to be imaged has been detected at the predetermined location along its trajectory.
It is appreciated that the triggering method of FIG. 15 is suitable not only for triggering imaging units as shown. The triggering method of FIG. 15 has more general applications such as triggering a momentary high illumination level of an article which is normally illuminated at a low level. For example, a suitable level of illumination for imaging fruit for analysis is approximately 3000 lux. A suitable level of illumination for imaging fruit merely in order to detect the time at which the fruit reaches a predetermined position along its trajectory is approximately 10 lux.
Alternatively, the method of FIG. 15 for triggering a high level of illumination or for triggering the imaging-for-analysis of the article in motion or for triggering any other system response may be eliminated. Instead, a time schedule for imaging articles may be implemented by a microswitch or by employing any other suitable device for physically sensing that the article in motion has reached a predetermined position along its trajectory, such as entry to enclosure 230 of FIG. 10. Microswitches must be delicate enough not to damage delicate articles such as fruit.
FIG. 16 is a pictorial illustration of a sample configuration for a spectral standard useful in conjunction with the apparatus of FIGS. 10-14, as described above with reference to FIG. 12. As explained above, each spectral standard is preferably utilized as an on-line internal reference to which each article may be compared, and thereby serves to standardize imaging over time by overcoming variations in optical conditions over time.
As shown, the spectral standard may comprise a plurality of patches having different spectral bands. The patches may serve as standards for intensity and/or dynamic range and/or spectral content. A sample plurality of materials which may be cut to size in order to form patches suitable for inspecting apples includes the following:
1608: A white opaque diffuser formed of Alumina or BaSO.sub.4.
1610: A neutral density filter having 3 db attenuation such as a GD-30-S.
1612: A red filter such as a a CA-600.
1614: A green filter such as a CA-550.
1616: An IR filter such as an LG-697.
1618: An IR filter such as an LG-790.
1620: A black scatterer or black paint such as Krylon black enamel.
All filters described above are commercially available from Corion Corp., MA, USA.
Each patch may be approximately square in configuration with suitable dimensions such as 7 mm.times.7 mm. However, the exact configuration of each patch is preferably not exactly square but rather curved so as to correspond to the curvature of the boundary of backgrounds 270 of FIG. 12.
It is appreciated that the particular spectral bands and arrangement of spectral patches specified hereinabove are merely exemplary and are not intended to be limiting. Selection of spectral patches is preferably determined by at least the type of article being graded, including variety of fruit if the article is a fruit, and the types of blemishes and color variations to be detected.
FIG. 17 is an optical schematic diagram, not to scale, of an individual camera unit 216 constructed and operative in accordance with a preferred embodiment of the present invention. As shown, each camera unit preferably comprises a monochromatic camera 320 such as a Model TC 655EC, commercially available from Burle Industries, Middlesex, England, an imaging lens 322, a spectral separation element 324 preferably including a prism as described in detail below, and an aperture 326.
Camera 320 preferably has a spectral response in the visible and near-IR portions of the spectrum and is actuated by external triggering in accordance with a suitable method such as the method described herein with reference to FIG. 15. If high intensity continuous illumination is employed, a camera with a short duration shutter may be employed.
Imaging lens 322 preferably comprises a high quality imaging lens with a relatively low F-number such as a P/N 71846, commercially available from JML Optical Industries, Inc., NY, USA, which has a focal length of 12.5 mm and an F-number of 1.3. Spectral separation element 324 is operative to provide a plurality of non-overlapping subimages of each article, each subimage having different spectral characteristics. A preferred embodiment of spectral separation element 324 is described in detail below with reference to FIG. 18.
Aperture 326 typically comprises a black panel having a central aperture whose diameter may be approximately 7 cm. Due to provision of aperture 326, the periphery of each subimage of the article is black, as best seen in FIG. 21. Therefore, the plurality of subimages of each article typically do not cover one another despite a small amount of overlap therebetween.
Reference is made briefly back to FIG. 10. Preferably, the optical path from the camera 320 to the imaging enclosure 230 of FIG. 10 is relatively long in order to provide a relatively small viewing angle such that a plurality of subimages of each article may be imaged. In order to provide a relatively long optical path without greatly enlarging the dimensions of housing 250, the optical path from camera 320 to enclosure 230 may be broken into path segments, such as path segments 330 and 332 of FIG. 10, which are optically associated with one another by means of mirrors (not shown).
FIGS. 18 and 19 are perspective and cross-sectional illustrations, respectively, of spectral splitting element 324 of FIG. 17, constructed and operative in accordance with a first preferred embodiment of the present invention. As shown, the spectral splitting element preferably has a pyramid configuration comprising a plurality of facets such as 4 facets 360, 362, 364 and 366. The spectral splitting element of FIGS. 18 and 19 is operative to provide 4 partially overlapping or nonoverlapping images.
The spectral splitting element may be similar in construction to photographic multiple image elements such as Filter 201 of the Cokin Creative Filters System, commercially available from Cromofilter S. A., RC, Paris, France, except that the pyramid need not be truncated at its top.
A plurality of filters generally corresponding in number to the number of pyramid facets is operatively associated with the pyramid facets. For example, two filters 380 and 382 are visible in FIG. 19 which are operatively associated with pyramid facets 360 and 364 respectively. Each filter may be attached by an adhesive or by any suitable mechanical means. The filters may be replaced or augmented by a dielectric or absorptive optical coating deposited on one or both surfaces 390 and 392 of each of the filters and/or deposited on one or more pyramid facets such as facets 360, 364 and 394.
According to one preferred embodiment of the present invention, each filter and/or optical coating defines a spectral waveband which substantially do not overlap with the spectral waveband of any of the other filters. A sample set of 4 spectral profiles for 4 filters corresponding to the four facets of the pyramid of FIG. 4, which filters are particularly suitable for inspecting certain varieties of apples, is illustrated in FIG. 28.
FIG. 20 is a perspective illustration of a spectral splitting element which is constructed and operative in accordance with an alternative embodiment of the present invention. The spectral splitter of FIG. 20 is similar to the spectral splitter of FIG. 18 except that 6 facets are provided instead of 4, thereby providing 6 images of an article, each having different spectral characteristics. More generally, any suitable number of facets may be employed, depending on the number of differently characterized images required in a particular application.
It is appreciated that the spectral splitting elements employed in the apparatus of the present invention need not be pyramidal in configuration, as shown in FIGS. 18-20 which are merely exemplary of suitable spectral splitting elements.
It is appreciated that spectral splitting elements such as the spectral splitting elements illustrated in FIGS. 18-20 may be useful in a wide variety of contexts other than imaging agricultural produce, particularly applications in which it is desired to capture an image of a fast moving article which is spectrally separated to a plurality of full relatively high resolution "subimages" thereof, as shown in FIG. 21.
FIG. 21 is a pictorial illustration of an image generated by imaging an apple using the apparatus of FIG. 18. As shown, the image comprises four non-overlapping subimages 400, 402, 404 and 406, each of which may have different spectral characteristics. For example, subimage 400 may be red, subimage 404 may be green, subimage 406 may be IR and the waveband of subimage 402 may be a waveband which corresponds to the wavelength of light projector 218 of FIG. 10 and which substantially does not overlap the wavebands of subimages 400, 404 and 406 as shown in detail below with reference to FIG. 28.
Providing an IR image in combination with at least one colored image is particularly advantageous for fruit inspection because blemishes may effectively be detected by comparing the IR image of the fruit to the colored images thereof.
It is appreciated that an image generated using the apparatus of FIG. 20 would include 6 subimages rather than 4 subimages as in FIG. 18.
Each subimage preferably comprises an image 408 of an apple and an image 410 of a spectral standard, such as the spectral standard of FIG. 16, which serves as an internal reference for evaluating image 408. In subimage 402, the imaged apple 408 may be illuminated via a grid, as explained above. The image 412 of the grid is preferably a mask comprising a plurality of alternating dark and bright stripes. For example, the bright stripes may be approximately one-quarter the width of the dark stripes.
As explained above, the four subimages are separated by a black region 414 due to provision of aperture 326 of FIG. 17.
FIG. 22 is an optical schematic diagram of a commercially available projector, the model 790 Xenon Arc lamp source, commercially available from Newport Corporation, California, USA, which may be employed to implement light projector 218 of FIGS. 10 and 13, with the following modifications:
Transmitting filters, such as an OG-580 and an RG-780, both commercially available from Schott, PA, USA, are placed adjacent the lamp assembly 498 and the projection lens 500 respectively. A grid slide comprising a black or opaque mask of stripes is placed adjacent the aperture plate 502.
Reference is now made to FIG. 23 which is a simplified flowchart of a preferred method for implementing image processing unit 18 of FIG. 1. FIG. 23 is described with reference to sorting and grading apples, specifically, however, it is appreciated that the method of FIG. 23 is generally appropriate for a wide variety of article inspection procedures. The method of FIG. 23, as specifically described below with reference to the Appendices, is suitable for an article inspection system which includes three imaging units, however, it is appreciated that the method may be suitably modified by the ordinarily skilled man of the art in order to apply to an article inspection system including any other number of cameras, such as four or more cameras.
The method of FIG. 23 preferably comprises the following steps:
Step 1808: Set-up data is loaded into the system. Setup data preferably includes the following items of information:
a. Data regarding the position, relative to the cameras' field of view, of the element if any which supports the article to be imaged, during imaging. For example, the set-up data may indicate whether or not cables supporting the article are visible by one of the cameras.
If support elements are visible, data is preferably provided which identifies the location of support elements. Preferably, each camera generates a binary mask for each subimage, using the following steps:
i. All pixels are compared to a suitable threshold value, such as a gray level value of 100, thereby to define a binary mask. For example, pixels exceeding the threshold may be marked with a 0 gray level value, and pixels falling below the threshold may be marked with a 255 gray level value.
ii. A dilation procedure is performed and then repeated on the binary mask in order to compensate for lateral movement of the cables or nets.
Any conventional method for binary image dilation may be employed.
b. Data regarding the spectral band and the location within the field of view of the camera, in camera coordinates, of each of a plurality of spectral patches to which each location of each article is compared. Suitable pluralities of spectral patches are described above with reference to FIG. 16.
c. Data identifying subimages to be analyzed. For example, the number of subimages to be analyzed may be specified as 4 and the 4 subimages may be identified as IR, red, green and grid subimages, where the grid subimage refers to the subimage in which the article to be imaged is illuminated through a grid. The first three subimages are collectively termed herein "the spectral subimages", and are used for size, shape, color and blemish categorizations. The subimage data preferably includes, for each camera, an indication of fixed offsets generated between subimages due to optical effects of the beam splitter.
Alternatively, the grid subimage may be omitted or may be replaced by an additional IR subimage. The two IR subimages, also termed herein the "IR band I" and "IR band II" subimages, are useful in detecting blemishes as described below with reference to step 1830 of FIG. 1830.
Step 1810: Optionally, profile data characterizing the variety of article to be inspected is loaded in image processing unit 18. For example, if the articles to be inspected are apples, substantially any variety of apple may preferably be processed, such as Red Delicious, Ida Red, Smith, Hermon, Golden Delicious, McIntosh and Jonathan. Sample data for a profile of a particular variety may include the following:
a. Definitions, preferably expressed as computational formulae, for each of a plurality of color gradings of a variety to be processed.
Color gradings may be based on accepted standards such as, for apples, the U.S.D.A. Standards for Grades of Apples.
b. Definitions, preferably comprising a fuzzy logic expert system rule base, for each of a plurality of types of blemishes which occur in a particular variety to be processed.
c. A computational formula for correcting the volume of the article, based on an assumption regarding the article in question, such as the assumption that the article is smoothly contoured or spherical. The correction preferably is based upon more accurate knowledge of the shape of a typical article of the particular variety in question, relative to the assumed geometric shape of articles which are not classified by variety.
A sample method for generating profile data for an undocumented type of article, such as a not-yet-profiled apple variety, is described below with reference to FIG. 26.
Step 1812: A plurality of digitized subimages is received from each of the plurality of cameras in electro-optical imaging unit 16 of FIG. 1. In the present example, it is assumed that 3 cameras are employed, each providing 4 subimages, termed herein the red, green, IR and grid subimages, giving a total of 12 subimages of each apple of which 9 are spectral subimages and 3 are grid subimages.
Step 1813: Since the four subimages are typically offset relative to one another, a suitable correction is preferably provided. Also, motion of the produce is compensated for.
Step 1814: Each of the spectral subimages is corrected to compensate for uneven illumination or shading of the imaged apple, using any suitable method. A suitable shading correction algorithm is described in the following publication, the disclosure of which is incorporated herein by reference:
Ballard, D. H. and Brown, C. M., Computer Vision, pp. 72-73, Prentice-Hall, Englewood Cliffs, N.J., 1982.
Appendix A, when used in conjunction with Appendix B, is a computer listing of a software implementation of a preferred method for implementing steps 1813 and 1814 of FIG. 23.
Step 1816: The boundaries of each of the 12 subimages are located.
Step 1816 may be performed only once per camera per article, preferably on the green subimage provided by each camera or on a subimage generated by averaging the green and red subimages provided by each camera. Once identified, the boundary of the green subimage or of the green/red subimage for each camera is then modified in order to generate the boundaries of the remaining subimages provided by the same camera. Typically, a predetermined x-y shift is sufficient in order to obtain the boundary of the red, grid or IR subimages from the boundary of the green subimage from the same camera.
As explained above, set-up step 1808 preferably includes the step of loading, for each camera, an indication of predetermined x-y shifts generated between subimages due to optical effects of the beam splitter.
A representation of the boundary, in suitable form such as an XY list or in vector form, is stored in memory.
Preferably, all steps from here on are performed only for pixels lying within the boundary as stored in memory, in order to eliminate unnecessary computations.
A suitable method for implementing step 1816 is described below with reference to FIG. 24 and Appendix C.
Step 1818: For each pixel in the image of the article, the color of that pixel is determined. For example, the pixels within each of the spectral subimages may be compared to a spectral standard such as the spectral standard of FIG. 16. A particular advantage of performing this step by comparison to a spectral standard is that on-line comparison with a spectral standard placed within the same optical chamber used to image the apples cancels out errors due to temporal variations in the illumination of the interior of the optical chamber and due to dirt.
Preferably, the spectral standard comprises a plurality of spectral patches, as illustrated in FIG. 16, each of which is assigned an identifying code. Each pixel within each of the spectral subimages is compared to the corresponding subimage of each of the plurality of spectral patches in order to identify the spectral patch whose subimage is closest in color or spectral characteristics to the subimage pixel. For example, if a brown color patch is provided, a bruise location on an apple will normally be found to be closest in color to the brown color patch. The subimage pixel is then assigned the identifying code of that patch.
A computer listing of a software implementation of a sample method for assigning identifying color codes to each of a plurality of locations in imaged StarKing apples which does not employ the method of comparison to a spectral standard is provided herein and is referenced Appendix D.
Step 1820: The grid subimage generated by each camera may be analyzed in order to either identify stem and calyx locations or to identify an absence of stem or calyx in the portion of the article viewed by the camera. Preferably, one or more of the spectral subimages are also analyzed in order to facilitate identification of stem and calyx. A preferred method for implementing step 1820 is explained in more detail below with reference to FIG. 25.
Optionally, if step 1836 below is included in the method of FIG. 23, step 1820 may be eliminated.
Step 1824: Each location along the apple surface is preferably assigned to exactly one of the three cameras before the fruit feature computation step described below is carried out.
According to one preferred embodiment of the present invention, predetermined portions of the article surface are assigned to each camera in accordance with a predetermined model of the shape of the article, such as a sphere.
A computer listing of a software implementation of a sample method for implementing step 1824, using a generally spherical model of the article, is provided herein and is referenced Appendix F. More specifically, the listing of Appendix F assumes that the cross sectional portion of the article corresponding to each line of the image is circular in shape.
Preferably, an image of at least a portion of the article is compared to the predetermined model of the shape of the article in order to determine discrepancies between the actual shape of the article and the predetermined model thereof. The determination of the predetermined portions of the article may then be corrected to take into account these discrepancies.
According to an alternative embodiment of the present invention, subimages generated by adjacent cameras are compared in order to identify registration therebetween, which typically defines areas of overlap therebetween. The overlap is taken into account such that each location is analyzed by taking into account only one image thereof generated by only one of the three cameras. Methods for determining the relative alignment of two subimages are described in the following publication, the disclosure of which is incorporated herein by reference:
Barnard, S. T. and Fischler, M. A., "Computational stereo", Computing Surveys, 14(4), 553-572, 1982.
Preferably, step 1824 comprises the step of identifying, for each of a majority of individual locations within an area of overlap between two camera units, the camera unit which images the individual location at the smallest angle to a normal to the surface of the article at the individual location.
Step 1828: The volume of the apple is computed, using any suitable method such as the following:
a. Each of the imaging units may compute, for each line 1 of the image generated thereby, the distance d.sub.1 between a "begin article" point B and an "end article" point E, as shown in FIG. 27.
b. For each line 1, a value D.sub.1 is computed as the product of the distance d.sub.1 and a constant factor F which compensates for the fact that the imaging unit generally images less than half of 64, the article. F may be computed by imaging a spherical article of known diameter and computing the ratio between the known diameter and the diameter of the imaged article.
c. For each imaging unit, an approximation to the volume of the apple in pixel units is computed, using the following formula:
d. For each imaging unit, the volume in pixel units computed in step c is converted to mm.sup.3 by multiplying the value generated in step c by a scaling factor.
e. The values generated in step c for each of the imaging units is averaged over imaging units to obtain an approximate value for the volume of the article in mm.sup.3.
A software implementation of steps a-d for a three camera system is provided herein in Appendix F.
In step 1830, an image is generated by combining the IR band I and IR band II images so as to obtain differences therebetween. The image may be generated by dividing the IR band I image by the IR band II image, or by subtracting the IR band I image from the IR band II image.
In step 1832, the image generated in step 1830 is inspected to identify blemish candidates which differ in appearance from the expected appearance of the apple. Specifically, contours for each discolored portion of the apple, also termed herein "blemish candidates", are identified.
A commercially available chip which is suitable for identifying contours is the L64290 chip, commercially available from LSI Logic Corporation, Milpitas, Calif., USA. Preferably, for each blemish candidate, features are recorded such as the area, perimeter, darkness inside the area, darkness outside the area and an aspect ratio, i.e. the ratio between the horizontal and vertical dimensions of the area.
Appendix G is a computer listing of a software implementation of a preferred method for implementing steps 1830 and 1832 of FIG. 23.
In step 1834, each blemish candidate is fuzzy-logic processed so as to classify it as one of a plurality of blemish types such as russeting, bruise and bitter pit.
Appendix H, taken together with Appendix I, is a computer listing of a software implementation of a preferred method for implementing step 1834 of FIG. 23.
Step 1836: The information regarding absence of or location of stem and calyx for each of the cameras, generated in step 1820, is combined over all cameras, in order to make a final identification of the locations of the stem and calyx. Preferably, the final identification of stem and calyx locations takes into account some or all of the following data:
a. A comparison of the green, red and IR outputs of the stem and calyx locations to a predetermined dark color which is known to be characteristic of stems and calyxes.
b. The shape of the fruit, such that an identification of a pair of stem and calyx locations which are not located along a central axis of the fruit, will be rejected.
It is appreciated that the information regarding the locations of the stem and calyx may be utilized in performing subsequent operations on the fruit, such as cutting, marking and packing the fruit.
Preferably, each of a plurality of pairs of blemish candidates is fuzzy-logic processed to identify one of the pairs of blemish candidates as including a stem and a calyx. Preferably, the results of step 1834 are employed to filter out blemish candidates which do not resemble stems and calyxes and therefore need not be fuzzy-logic processed.
If three cameras are employed, the set of blemish candidates, each pair of which is fuzzy-logic processed in step 1836, preferably includes two "dummy" blemish candidates positioned at the two blind spots of the imaging apparatus. Thereby, the system takes into account the possibility that the stem or calyx or both may be located at one or both of the two blind spots of the imaging apparatus.
Optionally, if step 1820 above is included in the method of FIG. 23, step 1836 may be eliminated.
Preferably, in step 1838, the classification of blemishes is reviewed and modified to take into account the stem/calyx position as determined by step 1836. For example, a blemish candidate which was classified in step 1834 as russeting may now be reclassified as a non-blemish if it surrounds the stem position.
As explained above, Appendix I includes a computer listing which, taken together with Appendix H, forms a computer listing of a software implementation of a preferred method for implementing step 1834 of FIG. 23. However, Appendix I also includes a computer listing of a software implementation of a preferred method for implementing steps 1836 and 1838 of FIG. 23.
In step 1840, the apples are graded based on the characteristics of the blemishes on each apple, as derived in steps 1832 and 1838.
Fuzzy logic technology is discussed generally in the following publication, the disclosure of which is hereby incorporated by reference:
Bezdek, J. C. and Sankar, K. Pal, Fuzzy models for pattern recognition, IEEE Pres 1992, ISBN 0-7803-0422-5.
Commercially available development packages include TILShell together with FCDS (Fuzzy C Development System), commercially available from Togay InfraLogic, Irvine, Calif., USA and the ADS 230 development system, commercially available from American NeuraLogix Inc., Sanford, Fla., USA.
It is appreciated that the steps of the flowcharts described in the present specification need not be performed in the illustrated order and specifically, the steps of the method of FIG. 23 need not be performed in the illustrated order. For example, step 1818 may be performed after step 1824 in order to save computing time. Also, steps 1820 and 1826 may be performed in parallel to steps 1818 and 1824 or after steps 1818 and 1824. Many other modifications of the illustrated order are possible.
Reference is made briefly to FIG. 24 which is a simplified flowchart of a preferred method for implementing step 1816 of FIG. 23. A computer listing of a software implementation of a sample method for implementing the method of FIG. 24 is provided herein and is referenced Appendix C.
Using the method of FIG. 24, the boundary of the apple may be found. In step 1850, at least one of the subimages of the apple is transformed into a binary image, comprising object pixels and background pixels, by a thresholding procedure. For example, if the pixels of the subimages are represented in 8 bits, a suitable threshold is a gray level of 200. Preferably, the green subimage is binarized because the green subimage has been found to differentiate well between the apple and its background.
In step 1852, a filtering procedure is employed to remove images of the support elements and artifactual images. The filtering procedure is operative to remove all portions of the image which are smaller than a selected filter.
According to one alternative embodiment of the present invention, filtering is initially carried out using a first annular disc defining a one-pixel wide ring and the filtering procedure includes a single step in which, for each location within the image for which the one-pixel wide ring is not entirely covered by a non-background entity, the entire portion of the image underlying the ring is removed.
If the width of the support elements supporting the agricultural produce is 5 pixels, a suitable radius for the first annular filter is, for example, 6 pixels.
Preferably, a second annular filter with a larger radius such as, for example, 20 pixels, is additionally employed in order to round off edges left by the small radius filter.
According to another alternative embodiment of the present invention, step 1852 is replaced by the step of employing a round disc filter. In this case, the filtering procedure includes two steps: an erosion step in which centers of portions which are smaller than the filter are removed; and a dilation step in which edges of large portions are restored, since these edges tend to be eroded in the erosion step. This embodiment may, for example, be implemented using a L64230 chip, commercially available from LSI Logic.
Typically, there is a tradeoff between the accuracy of the filtering operation which is achieved by employing a large filter, and between simplicity of computation which is achieved by employing a small filter.
In step 1854, the coordinates of the apple boundary, typically Cartesian coordinates, are identified. Typically, the algorithm positions itself on the top line of the binary subimage. If no object pixel is encountered in the top line, the algorithm advances to and through the next line, and so on, until on a particular line, the first object pixel is encountered. When this occurs, the border pixels on the left and on the right for that line are identified and the algorithm then advances to the next line until a line is found on which, again, there are no object pixels. At this point, the boundary coordinates are stored (step 1866).
Reference is now made to FIG. 25 which is a simplified flowchart of a preferred method, suitable for implementing step 1820 of FIG. 23, for analyzing a grid image generated by each camera in order to identify absence of or location of stem and calyx. The grid image may comprise a grid of bright lines superimposed on a dark background. Before employing the method of FIG. 25, the edges of the bright lines may be sharpened by using any suitable method such as application of an edge-enhancement operator to the image.
The method of step 1820 then generates, for each camera, an indication of a putative stem location or of an absence of stem, and an indication of a putative calyx location or of an absence of calyx. Locations are identified as putative stem or calyx locations if the curvature of the grid lines, superimposed on those locations, changes relatively sharply thereat.
A computer listing of a software implementation of a preferred method for implementing the method of FIG. 25 is appended hereto and is referenced Appendix E.
Reference is now made to FIG. 26 which is a simplified flowchart of a method whereby a human operator "teaches" the system about an individual type of article, such as an individual variety of agricultural produce such as an apple variety, and the system generates profile data for the apple variety.
The method of FIG. 26 preferably comprises the following steps:
Step 1900: The learning process preferably takes place with illumination and other optical conditions that resemble operative conditions as closely as possible.
Step 1902: Steps 1904, 1906 and 1908 are performed for each of a plurality of blemish types which it is desired to define.
Step 1904: An article of the appropriate variety, such as an apple, is imaged.
Step 1906: A human operator identifies the location of a blemish within the image of the apple, using suitable means such as a mouse.
Step 1908: The system analyzes the spectral characteristics of the location indicated in step 1906. For example, the system may compare the color/spectral value of each of a plurality of pixels within the location indicated in step 1906, to the spectral patches of a spectral standard such as the spectral standard of FIG. 16, and identify the spectral patch that best characterizes the type of blemish under consideration.
Steps 1910-1916: Each of a plurality of colors which characterize the apple variety, such as red, green and yellow, may be learned in steps similar to blemish learning steps 1902-1908 above.
Reference is now made to FIG. 28 which includes four graphs 1950, 1952, 1954 and 1956 of spectral characteristics of the four filters of FIG. 18 respectively, for applications in which a projected light grid is employed in conjunction with filter 1956. For applications in which a projected light grid is not employed, the fourth filter, instead of having the spectral characteristics of graph 1956, may have the spectral characteristics of graph 1957. According to one preferred embodiment of the present invention, the spectral profiles of the four filters are generally nonoverlapping. It is appreciated that the filters defined by the spectral characteristics of FIG. 28 are merely exemplary of suitable filters.
According to a preferred embodiment of the present invention, the four filters of FIG. 18 comprises a combination of filters, preferably including at least one IR filter, which is designed to facilitate distinction of bruised tissue from non-bruised tissue, as described below in detail with reference to FIG. 28. Reflectance spectra for bruised and non-bruised tissue are described in the following publication, the disclosure of which is incorporated herein by reference:
B. L. Upchurch et al, "Spectrophotometric study of bruises on whole, red delicious apples", Transactions of the ASAE, 33(2), March-April 1990.
Providing an IR filter is particularly advantageous for detecting blemished regions such as russeting, sun-burn, rot and bruising on apples.
FIG. 29 illustrates the conveying apparatus of FIG. 7 in operative association with the imaging apparatus of FIG. 10. It is appreciated that conveying apparatus constructed and operative in accordance with any of the embodiments shown and described herein may be similarly combined with imaging apparatus constructed and operative in accordance with any of the embodiments shown and described herein. A particular feature of the apparatus of FIG. 29 is that only the elements, if any, directly supporting the article to be inspected, such as the nets and cables shown and described above, are disposed interiorly of the imaging enclosure and therefore there is relatively little obscuring of the article.
Reference is now made to FIG. 30A which is a simplified side view illustration of singulating apparatus including a conveying junction 2098 comprising an interlaced pair of conveyors 2110 and 2160 suitable for conveying agricultural produce such as apples. Conveyors 2110 and 2160 may replace conveyors 170 and 148, respectively, of FIG. 7. Alternatively, conveyors 2110 and 2160 may replace conveyor 170, only, of FIG. 7.
The singulating apparatus of FIG. 30A receives non-singulated produce from a bulk conveyor 2100 which may comprise a conventional strip conveyor such as the strip conveyor commercially available from Matechet Rishon, Rishon Le Zion, Israel. The bulk conveyor 2100 feeds onto ascending singulating conveyor 2110. Preferably, means are providing adjacent the junction between conveyors 2100 and 2110 for urging apples from conveyor 2100 onto conveyor 2110, such as a brush 2120.
The singulator of FIG. 30A is also suitable for applications in which the ascending conveyor 2110 is partly submerged in water, and the non-singulated produce is provided within the water and is scooped up therefrom by ascending conveyor 2110, as explained above with reference to FIG. 3B.
The ascending conveyor 2110 typically comprises a plurality of rotating roller assemblies 2130 which are mounted on a guiding chain or belt 2140 in operative association with a plurality of guiding pulleys such as pulleys 2150, 2152 and 2154. Ascending conveyor 2110 is preferably inclined so as to allow apples to roll or slide backwards into spacings between the roller assemblies 2130. Rotation of the roller assemblies 2130 need not be linked with the motion of the guiding chain 2140.
Each roller assembly is configured for supporting an apple or other item of produce. Preferably, each roller assembly may be constructed as illustrated in detail in FIG. 30B.
Ascending conveyor 2110 interlaces with a cable conveyor 2160 including a plurality of cables such as 3 or 4 cables which may, for example, be similar to the cable conveyor shown and described above with reference to FIG. 5. The cables may be formed of any suitable material such as rubber or polystyrene. Cable conveyor 2160 is mounted on and driven on pulleys 2150, 2152, 2154, 2170 and 2180. In other words, cable conveyor 2160 and ascending conveyor 2110 have guiding pulleys in common.
Sample dimensions for the apparatus of FIG. 30A include: rollers: outer diameter--5 cm, width--6 cm, inter-roller spacing--8 cm; brush: diameter--25 cm; speed--30 rpm; length of ascending conveyor--50 cm; angle of ascending conveyor--13 degrees; diameter of pulleys--15 cm; apple conveying speed--5 apple/sec.
It is appreciated that singulation of the produce in the produce inspection system shown and described herein may occur at any suitable point. Conveyor 2110 of FIG. 30A may be, but need not be, a singulator. If conveyor 2110 is not a singulator, the dimensions of the rollers 2130 thereof may be, for example, approximately 2 cm, and the separation therebetween may be, for example, approximately 1 cm. Alternatively, rollers 2130 may be eliminated. In this case, singulation of the produce is provided at the junction 2098 between conveyors 2110 and 2160 by causing the velocity of conveyor 2160 to exceed the velocity of conveyor 2110. For example, the velocities of conveyors 2110 and 2160 may be 40 cm/sec and 150 cm/sec, respectively.
According to still a further alternative, singulation may not be performed by the apparatus of FIG. 30A. Instead, singulation may be performed by the apparatus of FIG. 32, as explained in detail below with reference to FIG. 32.
Reference is now made to FIG. 30B which illustrates an individual one of roller assemblies 2130, constructed and operative in accordance with one preferred embodiment of the present invention. Each roller assembly comprises a plurality of coaxial elements including article supporting discs 2170 defining a cradle 2172 for supporting an article such as an apple, cable supporting pulleys 2174 for supporting cables 2160 below cradle 2172, and, preferably, spacers 2176.
Article supporting discs 2170 may, for example, be formed of rubber and they may rotate at a suitable speed, such as 100 rpm, if it is desired to rotate the articles as they are conveyed. Cable supporting pulleys 2174 may, for example, be formed of a suitable low-friction metal or plastic. Spacers 2176 may, for example, be formed of a suitable metal or plastic.
It is appreciated that due to the configuration of FIG. 30B, the cables 2160 are below and spaced away from the apples being conveyed, until junction 2098 is reached, from which point the apples are supported by cables 2160.
FIGS. 31A-31C illustrate apparatus for selectably ejecting apples from a cable conveyor 2200 into a selected one of a plurality of bins 2310 corresponding to a plurality of categories into which the apples have been categorized. Typically, apple categorization and control of the ejecting apparatus of FIGS. 31A-31C is by means of an automatic unit such as data analysis/ sorting/grading unit 20 of FIG. 1.
The cable conveyor 2200 includes a plurality of cables such as 3, 4 or more cables. In the illustrated embodiment, the cable conveyor 2200 includes first, second and third cables 2210, 2220 and 2230. In FIG. 31A, the produce is moving from left to right, as indicated by an arrow 2250. In FIG. 31B, the produce is moving into the plane of the page. The rate at which the produce is conveyed is typically relatively high, such as 5 articles per second.
The produce ejecting apparatus of FIGS. 31A-31C includes, opposite each of bins 2310, at least one kicker 2240 and preferably a plurality of kickers, such as four kickers 2240, 2242, 2244 and 2246. Each kicker preferably includes a rotating disc 2260 which is arranged such that its tangential velocity substantially equals the velocity of the moving produce, as indicated by arrow 2250.
The magnitude of the velocity vector is preferably equal to the linear velocity of the produce, so as to prevent damage to the produce when engaged by the kicker. Disc 2260 may be formed of a soft rubber or plastic such as polyurethane. Sample dimensions for each disc 2260 are a diameter of 5 cm and a width of 2 cm. The separation between adjacent discs may be 6 cm center to center, or a 1 cm separation between perimeters.
Each kicker is actuated by suitable means such as a solenoid or pneumatic piston 2270 which causes the kicker to pass smoothly between two extreme positions: a first inactive, withdrawn position, such as the position of kicker 2240 of FIG. 31A, and a second active, protruding position, such as the position of kicker 2246 of FIGS. 31A and 31B. The amplitude of motion between the first and second positions is preferably 3 cm-5 cm.
When a kicker is in its inactive, withdrawn position, it does not interfere with the passage of an item of produce along the cables 2200. When a kicker is in its active, protruding position, it protrudes between cables 2220 and 2230 and engages a passing item of produce, acting to eject the item sideways, over cable 2210. Each kicker passes from its inactive position to its active position by extending along a direction indicated by an arrow 2280.
In FIG. 31A, a first apple 2290 is shown which is to be ejected and a second apple 2300 is shown, following the first apple, which is not to be ejected. Kicker 2240 is stationary, in its inactive position, since apple 2300 is not to be ejected. Kicker 2242 is withdrawing after ejecting apple 2290, as indicated by arrow 2302, since apple 2300 is not to be ejected. Kicker 2244 is in its active position, since apple 2290 is to be ejected, and kicker 2246 is moving into its active position, as indicated by arrow 2280, in anticipation of the arrival of apple 2290.
It is appreciated that, if desired, bins may be placed on both sides of the cable conveyer. In this case, kickers corresponding to the bins on one side of the cable conveyor, such as bin 2310, will protrude between cables 2220 and 2230 when active, whereas kickers corresponding to the bins on the other side of the cable conveyor will protrude between cables 2210 and 2230 when active.
FIG. 32 illustrates a preferred alternative to the active conveying apparatus of FIG. 7 which is similar to the apparatus of FIG. 7 except that a cushioning element and brushes are provided to facilitate the transition of apples from one conveying device to another. Elements in FIG. 32 which are similar to corresponding elements in FIG. 7 bear identical reference numbers for convenience.
In the illustrated embodiment, a cushioned landing pad 2190 is provided adjacent to and downstream of conveyor 170. The width W of the cushion 2190 is preferably relatively small, such as 2 cm, so that even articles which are not round tend to fall off the cushion 2190 toward a brush array comprising, in the illustrated embodiment, first and second brushes 2200 and 2210.
The first and second brushes 2200 and 2210 are provided upstream of cable conveyor 148, above and below cushioned landing pad 2190 or above and below a location slightly downstream of cushion 2190. A suitable separation between the brushes is, for example, 2 cm. The highest point of lower brush 2210 is typically slightly elevated relative to the plane defined by cable conveyors 148, such as at an elevation of 1 cm relative to the plane of conveyors 148. The linear velocity of the brushes is typically equal to the linear velocity of cables 148.
During operation, apples are deposited from loading conveyor 170 onto the cushioned landing pad 2190 from which they fall toward the brushes 2200 and 2210. The brushes facilitate transfer of the apples onto driven cables 148. Preferably, brushes 2200 and 2210 accelerate the top and bottom portions, respectively, of the apples to a horizontal velocity suitable for conveyance along cables 148, such as 1.2 m/sec. Brushes 2200 and 2210 preferably place the articles onto cables 148, rather than rolling them thereon, such that the separation between apples on cables 148 is maintained because the apples do not roll therealong.
It is appreciated that the brushes 2200 and 2210 may, alternatively, be replaced with functionally equivalent elements such as assemblies of soft plastic or rubber disks or sheets.
Preferably, a decelerating assembly 2220 is provided for decelerating apples arriving from cables 148 and facilitating the sequential transfer of these apples onto unloader 172. The decelerating assembly 2220 typically comprises a first brush 2230, a cable conveyor 2240, a pulley array 2250 for supporting and driving cable conveyor 2240, a second brush 2260 and a third brush 2270.
Cable conveyor 2240 may be approximately 60 cm long and comprise a plurality of rubber or polysterene cables which are typically thick relative to cables 148 and may, for example, have a cross-section of approximately 6 mm. The cables 2240 are associated with pulleys 2250 and are meshed or interlaced at their upstream end with the downstream end of cables 148 and at their downstream end with the upstream end of unloader 172. The association between adjacent conveyors may be similar to that illustrated in FIGS. 33A and 33B. The speed of the cable conveyor 2240 is preferably similar to the speed of unloader 172, such as approximately 0.7 m/sec. The separation between the tips of brushes 2230 and 2260 is typically approximately 2 cm.
The operation of the decelerating assembly 2220 is as follows: The brush 2230 lifts an apple riding off cables 148 onto cable conveyor 2240, also decelarating the bottom portion of the apple to the speed of the cable conveyor 2240.
Brush 2260 decelerates the upper portion of each apple so as to prevent it from rolling forward. In other words, the function of brush 2260 is substantially the reverse of the function of accelerating brush 2200.
Brush 2270 cushions the apple as it is transferred from conveyor 2240 to conveyor 172.
According to an alternative embodiment of the present invention, any of the various brushes in FIG. 32 may be omitted.
According to one alternative embodiment of the present invention, loading conveyor 170 conveys the articles in bulk and the articles are singulated in the course of being transferred to cables 148. For example, conveyor 170 may be a bulk conveyor, and a funnel may be provided upstream of cushioning pad 2190, such that the produce fall one at a time onto cushion 2190 and therefore are transferred one at a time onto cables 148. Separation between produce may be provided by having conveyor 148 move much faster than does conveyor 170. For example, the velocities of conveyors 148 and 170 may be 150 cm/sec and 40 cm/sec, respectively.
FIGS. 33A and 33B illustrate interlacing of cable 148 and associated pulleys 152; with cables 2240 and associated pulleys 2250. As shown, conveyor 148 is angled relative to conveyor 2240, for example at a 5 degree angle, so that the apples do not come into contact with the pulleys.
It is appreciated that a pair of meshed or interlaced cable conveyors such as those illustrated in FIGS. 33A-33B may be utilized more than once within a single produce sorting system and that one or more sequences of three, four or more pairs of interlaced cable conveyors may be provided within a single produce sorter.
A computer listing of a software implementation of a preferred method for implementing steps 1824 and 1828 of FIG. 23, is provided herein and is referenced Appendix F. The listing of Appendix F relies on a model of object contour according to which the object under inspection is smoothly contoured.
Appendix B includes a computer listing of a main program for an article inspecting system constructed and operative in accordance with a preferred embodiment of the present invention.
Appendix J includes computer listings of helper routines useful in conjunction with the other computer listings appended hereto, as explained in more detail below.
Appendix K includes computer listings of "include files" useful in conjunction with the other computer listings appended hereto, as explained in more detail below.
A suitable environment for running the listings appended hereto is a 486 AT with 16 megabytes of memory, an Imaging Technology CFG frame grabber board, a PCL 720 digital I/O board, commercially available from Advantech, Taipei, Taiwan, and a color screen, the following software products being installed on the 486 AT:
a. ITEX-CFG, commercially available from Imaging Technology, Boston, Mass., USA;
b. Fuzzy C Development System, commercially available from Togai Infralogic; and
c. Microsoft C compiler, Version 6.00. The frame grabber board may be configured as follows:
Base address--E0000, I/O address--240.
The PCL 720 board may be configured as follows:
I/O address--2A0.
The electronic components diagrammed in FIGS. 34A and 34B may be built and connected to the above commercially available components and to Cohu Model 4912 cameras, commercially available from Cohu, San Diego, Calif., USA, as shown in FIGS. 34A and 34B.
To load and operate any of the computer listings appended herewith, the following steps may be performed:
a. Generate a computer file of the listing. At the top of each listing in each Appendix, a title appears which should be the name of the computer file.
b. Use the batch file and DOS configuration file, a computer listing of which is appended herewith and is referenced Appendix L, to generate an executable program.
It is appreciated that the various appendices enclosed herewith are merely intended to provide an extremely detailed sample implementation of various preferred aspects of the present invention and are not intended to be limiting.
It is appreciated that features of the present invention that have been described in combination with other features may also be employed separately wherever suitable and that features which have been described separately may also be employed in combination with other features wherever suitable. For example, although the simultaneous inspection of generally all exposed areas of the inspected article is the preferred embodiment, alternatively, a plurality of portions of the exposed area of the inspected article may be inspected sequentially.
The present invention is operative in the context of apple inspection and has been described in relation to apple inspection. However, it is appreciated that the invention shown and described herein has a wide variety of applications in inspecting other articles such as but not limited to agricultural products generally and delicate agricultural products specifically, and also to round or spherical objects generally.
It will be appreciated by persons skilled in the art that the present invention is not limited to what has been particularly shown and described hereinabove. Rather, the scope of the present invention is defined only by the claims that follow. ##SPC1##
Claims
We claim:
1. Apparatus for inspecting articles comprising: a camera means operative to inspect generally all exposed surfaces of the article simultaneously; and an image processor for providing a reconstruction of the exposed surfaces of an article in which substantially every point on the exposed surfaces of the article is reconstructed exactly once.
2. Apparatus according to claim 1 wherein said camera means comprises at least first and second camera units and wherein said image processor comprises means for reconstructing predetermined at least first and second substantially nonoverlapping portions of the article in accordance with the images generated by the at least first and second camera units respectively.
3. Apparatus according to claim 2 wherein said predetermined portions of the article are determined in accordance with a predetermined model of the shape of the article, the model defining at least one model parameter.
4. Apparatus according to claim 3 and also comprising an image-model comparison unit for comparing an image of at least a portion of the article to the predetermined model of the shape of the article and for correcting the determination of the predetermined portions of the article to take into account discrepancies between the actual shape of the article and the predetermined model thereof.
5. Apparatus according to claim 1 wherein said camera means comprises at least first and second camera units and wherein said image processor comprises: apparatus for comparing first and second images of first and second respective portions of the article generated by the first and second camera units respectively, thereby to identify overlap between first and second images.
6. Apparatus according to claim 1 wherein the article is supported by an article supporting element disposed intermediate the article and said camera and wherein the image processor comprises an article supporting element identifier for differentiating an image of the article supporting element from the image of the article.
Patent Citations (30)
| Patent | Date | Inventor | Cited By |
|---|---|---|---|
| US2933613 | 1960-04-01 | Powers | |
| US3225892 | 1965-12-01 | Keesling | |
| US3867041 | 1975-02-01 | Brown et al. | |
| US3961701 | 1976-06-01 | Paddock et al. | |
| US4146135 | 1979-03-01 | Sarkar et al. | |
| US4204950 | 1980-05-01 | Burford, Jr. | |
| US4246098 | 1981-01-01 | Conway et al. | |
| US4515275 | 1985-05-01 | Mills et al. | |
| US4534470 | 1985-08-01 | Mills | |
| US4645080 | 1987-02-01 | Scopatz | |
| US4687107 | 1987-08-01 | Brown et al. | |
| US4735323 | 1988-04-01 | Okada et al. | |
| US4741042 | 1988-04-01 | Throop et al. | |
| US4799596 | 1989-01-01 | Mallant | |
| US4863041 | 1989-09-01 | Bailey | |
| US4878582 | 1989-11-01 | Codding | |
| US4940850 | 1990-07-01 | Satake | |
| US5010247 | 1991-04-01 | Smith et al. | |
| US5012116 | 1991-04-01 | Russell | |
| US5024530 | 1991-06-01 | Mende | |
| US5078258 | 1992-01-01 | van der Schoot | |
| US5085325 | 1992-02-01 | Jones et al. | |
| US5218555 | 1993-06-01 | Komai et al. | |
| US5315384 | 1994-05-01 | Heffington et al. | |
| US5339963 | 1994-08-01 | Tao | |
| US5390261 | 1995-02-01 | Huang et al. | |
| US5410637 | 1995-04-01 | Kern et al. | |
| US5526119 | 1996-06-01 | Blit et al. | |
| DEX2327355 | 1975-01-01 | ||
| WOXWO91/04803 | 1991-04-01 |
Non-Patent Literature (12)
- Ballard, D.H. and Brown, C.M., "Computer Vision", pp. 72-73, Prentice-Hall, Englewood Cliffs, New Jersey, 1982.
- Barnard, S.T. and Fischler, M.A., "Computational Stereo", Computing Surveys, 14(4), 553-572, 1982.
- Davenel et al., "Automatic Detection of Surface Defects on Fruit by Using a Vision System", J. Agricultural and Engineering Research, 41, 1-9, 1988.
- Graf, G.L., "Automatic Detection of Surface Blemishes on Apples Using Digital Processing", Cornell University, 1982.
- Hu, G. and Stockman, G., "3D Surface Solution Using Structured Light and Constraint Propagation", IEEE Trans. on PAMI, vol. 4, 390-402, 1989.
- Miller, B.K. and Delwiche, M.J., "A Color Vision System for Peach Grading", Trans. of the ASAE, vol. 32(4), 1484-1490, 1989.
- Rehkugler, G.E. and Throop, J.A., "Apple Sorting with Machine Vision", Trans. of ASAE, vol. 29(5), 1388-1397, 1986.
- Rehkugler, G.E. and Throop, J.A., "Image Processing Algorithm for Apple Defects Detection", Trans. of the ASAE, vol. 32(1), 267-272, 1989.
- Sarker, N. and Wolfe, R.R., "Computer Vision Based System for Quality Separation of Fresh Market Tomatoes", Trans. of the ASAE, vol. 28, 1714-1718, 1985.
- Shrikhande, N. and Stockman, G., "Surface Orientation From a Projected Grid", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 11(6), Jun. 1989.
- Upchurch, B.L. et al., "Spectrophotometric Study of Bruises on Whole, Red Delicious Apples", Trans. of ASAE, 33(2), Mar.-Apr. 1990.
- Wolf, R.R. and Sandler, W.E., "An Algorithm for Stem Detection Using Digital Image Processing", Trans. of the ASAE, 28, 641-644, 1984.