US 5,343,538 AGrant
Method and an Apparatus for Identifying an Object Using Quantile Partitions
Issue Date:1994-08-30
•48 Claims
•3 Drawing Sheets
Abstract
The present invention relates to a method and an apparatus for identifying an object under examination. The object has a discernible boundary. An image of the object is first formed. The image is segmented to form a plurality of pixels. The boundary of the object under examination in the image is then detected. The parameters of the pixels measured within the boundary detected are formed in an ordered sequence: where P.sub.1 is the first value of the parameter of the pixel within the boundary detected; where P.sub.N is the last value of the parameter of the pixel within the boundary detected; N is the total number of pixels within the boundary detected; and k is the index to the N pixels. The P.sub.k data point is determined such that k, the index satisfies the following relationship: ##EQU1## M--total number of quantile partitions i-i.sup.th fraction of partition or i.sup.th quantile partition. The object under examination is identified based upon the P.sub.k data point value, which is the ith quantile in M partitions developed from the above relationship associated with the object.
Metadata
Assignee
- International Remote Imaging Systems, Inc.
Inventor
- Harvey L. Kasdan
Application Information
Application Number:US 9560564
Filing Date:1992-10-02
Priority Date:1992-10-02
Art Unit:266
Classifications
IPC:
G06K 946
Field of Search:
38221;22;6;54;25;18;17;51
Patent Drawings (3 sheets)
Description
A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
1. Technical Field
The present invention relates to a method and an apparatus for identifying an object under examination, wherein an image of the object is formed. More particularly, the present invention relates to a method and an apparatus for identifying an object based upon the object having a discernable boundary and a novel feature termed "quantile".
2. Background Of The Invention
Methods and apparatuses for identifying an object are well-known in the art. The art of identifying biological samples is replete with various techniques for identifying the type of biological samples under examination. See, for example, U.S. Pat. No. 4,175,860. Formation of histograms of various parameters are also known. See for example U.S. Pat. No. 3,851,156.
Heretofore, one of the known methods for identifying an object is to use the size of the object. Other parameters of the object which can be used to identify an object include color, internal optical density, and average intensity within the boundary of the object.
In some cases, however, it is not possible to identify an object uniquely or to differentiate two objects if certain of their parameters yield the same result. Thus, for example, if two objects have the same size and have the same internal optical density and have the same average visible light intensity, then it is not possible to identify one of the objects or to differentiate these two objects, based upon these parameters. In addition, some of these parameters are computationally intensive requiring considerable amount of time or computer resources. Thus, the present invention contemplates a novel method and apparatus for identifying an object or to differentiate two objects based upon a new parameter which is termed "quantile".
Summary of the Invention
In the present invention, a method of identifying an object, under examination, which has a discernible boundary is disclosed. The method comprises the steps of forming an image of the object. The image is segmented to form a plurality of pixels. The boundary of the object under examination in the image is then detected. A measurement is made of a parameter of each pixel within the boundary detected. The parameter values of the pixels measured within the boundary detected are formed in an ordered sequence. The particular measured parameter value P.sub.k, where k is an index of the ordered sequence, is determined such that the index k satisfies the following relationship: ##EQU2## N--is the total number of pixels;
M--is the total number of quantile partitions
i--index corresponding to the i.sup.th fraction of partition or i.sup.th quantile
The object under examination is then identified based upon the Pk parameter value which is the i.sup.th quantile within M partitions developed from the above relationship associated with the object.
The present invention also contemplates an apparatus to carry out the foregoing method.
Brief Description of the Drawings
FIG. 1 is a schematic graphical illustration of two objects "a" and "b" which have the same size and shape and the same average intensity but which are different.
FIGS. 2a and 2b (FIG. 2) are a graph showing a histogram of intensity of light versus the number of pixels in the boundary detected.
FIGS. 3a and 3b (FIG. 3) are a graph showing a cumulative histogram of intensity of light versus the cumulative number of pixels in the boundary detected, and with the quantile feature determined for each object.
FIG. 4 is a table of pre-determined relationships or quantiles of various known objects.
FIG. 5 is an ordered sequence of the values of the parameter measured for the pixels in the boundary.
FIG. 6 is a block diagram of an apparatus suitable for carrying out the method of the present invention.
Detailed Description of the Drawings
Referring to FIG. 1, there is shown a schematic diagram of an image 10 of two objects: object "A" 12 and "B" 14. Although the description set forth hereinafter is with reference to the method of the present invention to distinguish object "A" from object "B", it can be seen that the method of the present invention is equally applicable to identifying a single object which is under examination.
Each of the objects 12 and 14 has a discernible boundary 18 and 20, respectively. As shown in FIG. 1, each of the objects 12 and 14 has the same size and shape. Although for illustration purposes, objects 12 and 14 are shown as having the same size and shape, it is not necessary to the practice of the present invention. The purpose of illustrating the method of the present invention using objects 12 and 14 having the same size and shape is to show how the method of the present invention can be used to distinguish objects 12 and 14 under examination when other parameters, such as size and shape, cannot be used.
Objects 12 and 14, of course, represent images of real objects, such as cells, genes or other biological or non-biological objects. In the method of the present invention, an image 10 is taken of the objects 12 and 14. The image 10 can be formed by using an electronic video camera, such as a CCD (the specific details of an apparatus suitable to carry out the method of the present invention will be discussed hereinafter). The image 10 is then filtered in accordance with the different color filters that can be used to distinguish the type of color of the object. Each of the different color images is segmented, to form a plurality of pixels. Each pixel is then digitized. The boundary of each of the objects 12 and 14 is determined. This can be done, for example, by the method disclosed in U.S. Pat. No. 4,538,299, which is incorporated herein by reference.
Once the boundary of each of the objects 12 and 14 is determined, one of the parameters of each pixel inside the boundary is measured. One parameter can be the intensity of visible light. Thus, the intensity of visible light at each pixel, positioned inside the boundary positions 18 and 20, is measured.
One prior art method is to form a histogram of intensity versus the number of pixels. Referring to FIG. 2, there are shown two histograms. The histogram of FIG. 2a corresponds to the object 12 whereas the histogram of FIG. 2b corresponds to the object 14. As can be seen from FIG. 1, object 12 has a substantially uniform intensity (the dark image indicates brightness) throughout the entire region within the boundary 18. Thus, the histogram, as shown in FIG. 2a, shows substantially N.sub.1 pixels each having an intensity of I.sub.1. In contrast, object 14 has a small area 16 that has greater intensity than the uniform intensity I.sub.1 of object 12. The histogram, as shown in FIG. 2b, shows that Nz number of pixels in the spot 16 have substantially an intensity value of I.sub.2, with N.sub.1 >N.sub.2 and I.sub.2 >I.
Once the histogram of each object 12 and 14 has been formed, a cumulative histogram is then developed. The cumulative histogram is formed by summing the number of pixels that has at least a certain intensity value. If N=f(I), where N is the number of pixels having an intensity level I, then Cum N=F(I) where Cum N is the cumulative number of pixels having an intensity level.ltoreq.I, and where ##EQU3## The cumulative histograms of the histograms shown in FIG. 2a and 2b, are shown in FIG. 3.
The particular parameter value, such as intensity of visible light, associated with a cumulative fraction of all pixels, is termed a quantile. Thus, from FIG. 3, the 50th quantile is I.sub.3 for object 12 and I.sub.4 for object 14. It should be noted that unlike conventional percentile representations, which are even increments of discrete percents, quantile representations are not limited to discrete, even increments. Thus, an object may be identified by its ith, jth, or kth quantile, where i, j, and k do not necessarily occur in discrete, even increments.
Since the quantile relationship, shown in FIG. 3, associated with each objects 12 and 14 is different, the objects 12 and 14 can be distinguished from one another, based upon this developed quantile relationship. Alternatively, to uniquely identify a particular object, such as object 14, the quantile relationship associated with the object 14 can be compared to a table of pre-determined relationships of other known objects whose quantile relationships have been pre-determined. The table of pre-determined quantile relationships of other known objects can be based upon experimental results. The comparison of the quantile relationship to the table of predetermined quantile relationships would then serve to identify uniquely the type of the object that is under examination.
For the purpose of distinguishing the objects 12 and 14 from one another or to identify an unknown object by comparing the quantile relationship, it is not necessary to compare each ordered pair of numbers (Q,I) (where Q is the quantile number and I is intensity associated therewith), to each other (in the case of two particles) or to a table (in the case of attempting to uniquely identify an object). One method is to compare the associated intensity value of the object under examination at a particular quantile number to the associated intensity value of another object at the same quantile number, or to a table of associated intensity values of identified objects at the same quantile number. Thus, as shown in FIG. 3, the 50th quantile for the object 12 has an intensity value of I.sub.3 and the same 50th quantile for the object 14 has the value I.sub.4. These intensity values can be compared to each other to distinguish one object from another. Alternatively, the 50th quantile of one of the objects can be compared to a table of intensity values of identified particles whose 50th quantile have been pre-determined. See FIG. 4. By comparing the 50th quantile of the unknown object to the table, the unknown object can be identified.
Of course, if we are dealing with objects of biological particles having statistical variations in intensity, it may not be possible to determine precisely the intensity associated with a particular quantile for the same particle under all cases of examination. Thus, a table relationship may encompass a range of values of intensity for a particular quantile such that the object having that range of intensity values, can be uniquely identified. The range of values can be developed from representative samples identified from experimental results.
The foregoing example illustrates a method of distinguishing objects wherein the average intensity of the two objects 12 and 14 are indistinguishable. However, using the intensity of visible light and the quantiles developed therefor, the objects can be distinguished from one another. Parameters, other than intensity of light can also be used to distinguish the particles. Thus, another parameter suitable for use is differentiating by color representation. Color can be represented by three primary colors of red, blue, and green. Alternatively, color can also be represented by hue, intensity, and saturation; or by cyan, magenta and yellow.
In one particular embodiment, the difference in color representation of the primary colors is used. The parameter of log(a)-log(b) may be employed, where a, and b are intensities of red and blue primary colors, respectively. Other possible combinations include: a being green and b being red; or a being green and b being blue.
The problem with the foregoing method of calculating quantiles is that a cumulative histogram must be formed. This necessitates the creation of a plurality of discrete "bins". A bin is defined by a cumulative fraction of all the pixels from a first fraction to a second fraction. As can be seen from FIG. 3, the entire value of the cumulative histogram from 0 to IT must be divided into a number of discrete bins. The i.sup.th quantile is the value of I corresponding to the i.sup.th cumulative fraction of the pixels. For parameters, such as log(a)-log(b), as mentioned hereinabove, the range of values can be enormous (due to the nature of the logarithmic operation). Thus, the choice of the boundary of the bins may become difficult.
In the method of the present invention, quantiles can be calculated without calculating a cumulative histogram. In the method of the present invention, the values of the parameters measured for each of the pixels within the boundary detected are placed in an ordered sequence. This is shown in FIG. 5. As previously discussed, for the example used in the method of the present invention, the objects 12 and 14 are assumed to be the same size and shape. Thus, each of those objects 12 and 14 would be segmented into the same number of total pixels, N. The parameter, for example of intensity of visible light, for each of the pixels within the boundary is then placed in an ordered sequence. P.sub.1 is the value of the pixel having the lowest intensity of visible light. P.sub.N is the value of the pixel having the largest value of the parameter, such as intensity visible light. The values of the parameter of the pixels measured are then formed in an ordered sequence. A value of P.sub.k represents the parameter of the pixel between the smallest value and the largest value with k being the index to the N pixels.
Thereafter, k is chosen such that k satisfies the following relationship: ##EQU4## where M is the total number of quantile partitions and i is the i.sup.th fraction of partition or i.sup.th quantile. For the i.sup.th chosen quantile, once k is determined, the intensity value of the data point P.sub.k would be the intensity value associated with the i.sup.th quantile, or simply the value of the ith quantile. Thereafter, the object under examination can be distinguished from other objects based upon the i.sup.th quantile.
Alternatively, the object under examination can be distinguished from all other objects based upon a table of predetermined values of corresponding i.sup.th quantile. This can be accomplished even where the two objects do not have the same number of pixels. Thus, for example, object 1 has N pixels and Object 2 has N.sub.2 pixels. The parameters of the pixels are placed in an ordered sequence as follows:
where k is the index. For each object, the P.sub.k data point is determined such that k satisfies the following: ##EQU5## where for Objects 1 and 2 i and M are chosen to be the same. The resultant value of P.sub.k for each object is compared to one another to differentiate the objects. Of course, as previously discussed, Objects 1 and 2 can be the same type of object even if P.sub.k1 does not equal P.sub.k2. There can be a range of values for a particular quantile such that objects having values within that range will still be the same type.
As can be seen from the foregoing, with the method of the present invention, no cumulative histogram calculation is made. As a result, there is no need to set the size of the bin to calculate the cumulative histogram. A simple comparative relationship to determine the index of the data points is determined from which the value of the data point of the parameter measured is then obtained which corresponds to the quantile of interest.
An example of the method of the present invention can be seen with reference to the following Table 1, with an example of three particles: Particle 1, Particle 2, and Particle 3, each having 50 pixels of varying intensity of the same parameter.
The next table, Table 2, shows various statistical values computed for the three particles.
The next table, Table 3, shows the creation of a plurality (10) of "bins" with each bin containing a count of the number of pixels having a value between a first count to a second count. Thus, Bin #1 contains all the pixels having intensity values between 0 and 10. For particle #1, there would be 13 such pixels. For particles #2 and #3, there would be 40 and 40 respectively. The 10 bins are chosen to partition the maximum range of pixel intensity values into 10 fractions. Since the maximum range of intensity value is zero to 41.23, 97.00, and 97.00 for particles #1, #2 and #3 respectively, the 10 bins are chosen to delineate pixel intensity values of 0-10-20- . . . -90-100. The column entitled Cumulative Histogram is a cumulative count (shown as a %) of the total number of pixels having an intensity value between zero and the value associated with the Bin #. As can be seen from Table 3, particles #2 and #3 are indistinguishable based upon cumulative histogram. Further, as previously discussed, for intensity values based upon log(), the value can be large. Thus, the range of pixel intensity values will be large, rendering difficult the division of the intensity values into a manageable number of bins.
The following table, Table 4, shows the table of pixel intensity values (Table 1) rearranged in accordance with the method of the present invention. The pixel intensity values are ordered in a sequential order, from highest to lowest. (Of course, the ordered sequence can be from the lowest to the highest.) In this example, P.sub.50 for each of the three particles have the values 6.46, 2.87 and 1.66, respectively, while P.sub.1 is 41.23, 97.00 and 97.00 respectively.
where Rank is the order in sequence by the intensity of the pixel, Pixel# is the number of the pixel from Table 1, Intensity is the value of the intensity for the corresponding Pixel# from Table 1, and Percent is the percent of particles that would have intensity value shown under Intensity or less.
Thus, the pixel having the Rank of 3 for Particle is Pixel# 45 from Table 1. Pixel# 45 for Particle 1 has an intensity value of 34.80. Further 95.91% of all pixels in Particle 1 have intensity values of 34.80 or less.
If the number of "bins" or fractions, M is chosen to be 10 (the same as the example based upon a cumulative histogram calculation), and N-number of pixels=50, then ##EQU6## where i is the i.sup.th bin. If i=10 ##EQU7## k.sub.10 =46, 47, 48, 49, 50.
If we choose k.sub.10 =50, then P.sub.k10 =6.46, 2.87, 1.66 for particles #1, #2 and #3, respectively.
Since P.sub.k10 =2.87 for particle #2 clearly differs from P.sub.k10 =1.66 for particle #3, particles #2 and #3 can be differentiated, while other statistical based analysis such as histogram, and cumulative histogram are unable to differentiate the particles. Furthermore, although a single value was compared, clearly a range of values can be compared. Note that the same holds true for P.sub.k9 (K.sub.9 =45)=8.55, 3.77, 1.98 and P.sub.k9 (K.sub.8 =40)=9.16, 4.25, 2.25 for particles #1, #2 and #3 respectfully.
An apparatus 30 suitable for carrying out the method of the present invention is shown in FIG. 6. The apparatus 30 comprises a microscope 32, which is focused on an examination area 33. The examination area can be a microscopic slide or a flow cell such as that disclosed in U.S. Pat. No. 4,338,024. A camera 34 is attached to the microscope 32 and is adapted to image a portion of a suspension having a particle therein. The camera 34 is preferably of a raster scan type and may be a CCD camera model XC-711 manufactured by Sony. The camera 34 can form an electrical image of the field of view as seen through the microscope 32. Further, the camera 34 segments the image into a plurality of pixels, with an electrical signal corresponding to each pixel of the image. The camera 34 also outputs a plurality of signals (3) one for each of the colors (Red, Blue, and Green).
Each of the signals from the camera 34 is supplied to a digitizer 36, which digitizes the image intensity of each pixel into an electrical signal representing the grey scale value. Preferably, the pixel is digitized into a gray scale value between 0 and 255, inclusive.
From the digitizer, the digitized gray scale value is supplied to a memory 38, where the values are stored. The memory 36 can be a RAM. A boundary detector 40, such as that disclosed in U.S. Pat. No. 4,538,299 operates on the image in the memory 38 and detects the boundary of the image detected. The result is also stored in the memory 38.
From the memory 38, a computer 42 operates on the image and measures a parameter, such as intensity of light, of each pixel within the boundary detected.
In the event the apparatus 30 seeks to determine the identity of a single particle under examination, then the particular quantile value derived by the computer 42 is compared to the table of stored values 46 of previously identified particles. The table 46 can be a ROM or another memory. The result can be displayed on the display 44.
In the event the apparatus 30 seeks to differentiate a plurality of objects under examination, then a quantile is developed for the same parameter for each of the pixels in the boundary of each of the images of the different objects. From the quantile, the particles under examination can be differentiated from one another. The result can be displayed on the display 44.
The computer 42 carries out its tasks in accordance with a computer program. A copy of a program to compute quantiles written in the C language for execution under the MS-DOS operating system is as follows: ##SPC1##
Claims
I claim:
1. A method of identifying an object under examination, said object has a discernible boundary; said method comprising the steps of: a) forming an image of said object; b) segmenting said image to form a plurality of pixels; c) detecting the boundary of said object under examination in said image; d) measuring a parameter, P, for each pixel inside the boundary detected, resulting in a plurality of values for the parameter P measured; e) forming an ordered sequence, having a form P.sub.1. . . P.sub.k. . . P.sub.N wherein P.sub.1 represents a first value of the ordered sequence of measured parameter P; P.sub.N represents a last value of the ordered sequence of measured parameter P where N represents a total number of pixels inside the boundary detected; P.sub.k represents a kth value of the ordered sequence of measured parameter P having a value between P.sub.1 and P.sub.N, where k represents an index to the N pixels; f) determining one or more values of k from a relationship of the form, ##EQU8## where M represents a total number of quantile partitions for the measured parameter P; represents an i.sup.th fraction of partition or i.sup.th quantile partition for the measured parameter P; g) determining the value of P.sub.k from the one or more values of k determined in step (f); and h) identifying said object under examination, based upon the value(s) of P.sub.k determined from step associated with said object.
2. The method of claim 1 wherein P.sub.1 is the smallest value of P.sub.N is the largest value.
3. The method of claim 1, wherein said identifying step further comprising: comparing the value(s) of P.sub.k determined from step (g) for said object under examination, to a table of pre-determined relationships of other identified objects to identify the object under examination.
4. The method of claim 3, wherein said table of pre-determined relationships of other identified objects comprises a range of values.
5. The method of claim 3, wherein said table of predetermined relationships of other identified objects comprises a single value.
6. The method of claim 4, wherein said table of pre-determined relationships are based upon experimental results.
7. The method of claim 6, wherein said table of predetermined relationships of other identified objects comprises a single value.
8. The method of claim 1, wherein said parameter is intensity of visible light.
9. The method of claim 1, wherein said parameter is a function of color representations at the same pixel location.
10. The method of claim 9, wherein said color representation is based upon three primary colors.
11. The method of claim 9, wherein said color representation is based upon hue, intensity, and saturation of colors.
12. The method of claim 10, wherein said parameter is wherein a and b are the intensities of two different colors.
13. The method of claim 12 wherein a is red color and b is blue color.
14. The method of claim 12, wherein a is blue color and b is green color.
15. The method of claim 12, wherein a is green color and b is red color.
16. A method of differentiating a plurality of different objects (object 1 and object 2) under examination, wherein each object has a discernible boundary; said method comprising the steps of: a) forming a plurality of images, with an image of each of said plurality of objects; b) segmenting each of said plurality of images to form a plurality of pixels; c) detecting the boundary of each object under examination in each of said images; d) measuring a parameter, P, for each pixel inside the boundary of each object detected, resulting in a plurality of values for the parameter P measured for each object; e) forming an ordered sequence having a form P.sub.11. . . P.sub.k1. . . P.sub.N1 for object 1, wherein P.sub.11 represents a first value of the ordered sequence of measured parameter P for object 1; P.sub.N1 represent a last value of the ordered sequence of measured parameter P for object 1 where N1 represents a total number of pixels inside the boundary of object 1 detected; P.sub.k1 represents a kth value of the ordered sequence of measured parameter P having a value between P.sub.11 and P.sub.N1, wherein k represents an index to the N1 pixels; and an ordered sequence having a form P.sub.12. . . P.sub.k2. . . P.sub.N2 for object 2, wherein P.sub.12 represents a first value of the ordered sequence of measured parameter P for object 2; P.sub.N2 represents a last value of the ordered sequence of measured parameter P for object 2 wherein N2 represents a total number of pixels inside the boundary of object 2 detected; P.sub.k2 represents a kth value of the ordered sequence of measured parameter P having a value between P.sub.12 and P.sub.N2, where k represents an index to the N2 pixels; f) determining, for each ordered sequence formed, one or more values of k from a relationship of the form ##EQU9## wherein M represents a total number of quantile partitions for the measured parameter P; i represents i.sup.th fraction of partition or i.sup.th quantile partition for the measured parameter P; g) determining the value(s) of P.sub.k, for each object, from the one or more values of k determined in step (f) for identical values of i and M; h) comparing the P.sub.k value(s) determined in step (g) for each object to one another to differentiate the objects.
17. The method of claim 16, wherein said parameter is intensity of visible light.
18. The method of claim 16, wherein said parameter is a function of color representations at the same pixel location.
19. The method of claim 18, wherein said color representation is based upon three primary colors.
20. The method of claim 18, wherein said color representation is based upon hue, intensity, and saturation of colors.
21. The method of claim 19, wherein said parameter is where a and b are the intensities of two different colors.
22. The method of claim 21 wherein a is red color and b is blue color.
23. The method of claim 21, wherein a is blue color and b is green color.
24. The method of claim 21, wherein a is green color and b is red color.
25. An apparatus for identifying an object under examination, said object has a discernible boundary; said apparatus comprising: a) means for forming an image of said object; b) means for segmenting said image to form a plurality of pixels; c) means for detecting the boundary of said image of said object under examination; d) means for measuring a parameter, P, for each pixel inside the boundary detected, resulting in a plurality of values for the parameter P measured; e) means for forming an ordered sequence having a form P.sub.1. . . P.sub.k. . . P.sub.N wherein P.sub.1 represents a first value of the ordered sequence of measured parameter P; P.sub.N represents a last value of the ordered sequence of measured parameter P where N represents a total number of pixels inside the boundary detected; P.sub.k represents a kth value of the ordered sequence of measured parameter P having a value between P.sub.1 and P.sub.N, wherein k represents an index to the N pixels; f) means for determining one or more values of k from a relationship of the form, ##EQU10## where M represents a total number of quantile partitions for the measured parameter P; i represents an i.sup.th fraction of partition or i.sup.th quantile partition for the measured parameter P; g) means for determining the value(s) of P.sub.k from the one or more values of k determined in step (f); and h) means for identifying said object under examination, based upon the value(s) of P.sub.k determined from step (g) associated with said object.
26. The apparatus of claim 25 wherein P.sub.1 is the smallest value and P.sub.N is the largest value.
27. The apparatus of claim 25, wherein said means for identifying further comprising: means for comparing the value(s) of P.sub.k determined from step (g) for said object under examination, to a table of pre-determined relationships of other identified objects to identify the object under examination.
28. The apparatus of claim 27, wherein said table of pre-determined relationships of other identified objects comprises a range of values.
29. The apparatus of claim 26, wherein said parameter is intensity of visible light.
30. The apparatus of claim 29, wherein said parameter is a difference of color representations at the same pixel location.
31. The apparatus of claim 30, wherein said color representation is based upon three primary colors.
32. The apparatus of claim 30, wherein said color representation is based upon hue, intensity, and saturation of colors.
33. The apparatus of claim 31, wherein said parameter is where a and b are the intensities of two different primary colors.
34. The apparatus of claim 33, wherein a is red color and b is blue color.
35. The apparatus of claim 33, wherein a is blue color and b is green color.
36. The apparatus of claim 33, wherein a is green color and b is red color.
37. An apparatus for differentiating a plurality of objects (object 1 and object 2) under examination, each of said objects has a discernible boundary; said apparatus comprising: a) means for forming an image of each of said plurality of objects to form a plurality of images; b) means for segmenting each of said plurality of images to form a plurality of pixels; c) means for detecting the boundary of each of said plurality of objects under examination in each of said plurality of images; d) means for measuring a parameter, P, for each pixel inside the boundary of each object detected, resulting in a plurality of values for the parameter P measured for each object; e) means for forming an ordered sequence having a form P.sub.11. . . P.sub.k1. . . P.sub.N1 for object 1, wherein P.sub.11 represents a first value of the ordered sequence of measured parameter P for object 1; P.sub.N1 represents a last value of the ordered sequence of measured parameter P for object 1 wherein N1 represents a total number of pixels inside the boundary of object 1 detected; P.sub.k1 represents a kth value of the ordered sequence of measured parameter P having a value between P.sub.11 and P.sub.N1, where k represents an index to the N1 pixels; and an ordered sequence having a form P.sub.12. . . P.sub.k2. . . P.sub.N2 for object 2, wherein P.sub.12 represents a first value of the ordered sequence of measured parameter P for object 2; P.sub.N2 represents a last value of the ordered sequence of measured parameter P for object 2 wherein N2 represents a total number of pixels inside the boundary of object 2 detected; P.sub.k2 represents a kth value of the ordered sequence of measured parameter P having a value between P.sub.12 and P.sub.N2, where k represents an index to the N2 pixels; f) means for determining, for each ordered sequence formed, one or more values of k from a relationship of the form ##EQU11## where M represents a total number of quantile partitions for the measured parameter P; i represents i.sup.th fraction of partition or i.sup.th quantile partition for the measured parameter P; g) means for determining, for each object, the value(s) of P.sub.k from the one or more values of k determined in step (f) for identical values of i and M; h) means for comparing the P.sub.k value(s) determined in step (g) for each object to one another to differentiate the objects.
38. The apparatus of claim 37 wherein P.sub.1 is the smallest value and P.sub.N is the largest value.
39. The apparatus of claim 37, wherein said means for identifying further comprising: means for comparing the value(s) of P.sub.k determined from step (g) for said objects under examination, to a table of pre-determined relationships of other identified objects to identify the objects under examination.
40. The apparatus of claim 39, wherein said table of pre-determined relationships of other identified objects comprises a range of values.
41. The apparatus of claim 38, wherein said parameter is intensity of visible light.
42. The apparatus of claim 41, wherein said parameter is a difference of color representations at the same pixel location.
43. The apparatus of claim 42, wherein said color representation is based upon three primary colors.
44. The apparatus of claim 42, wherein said color representation is based upon hue, intensity, and saturation of colors.
45. The apparatus of claim 43, wherein said parameter is where a and b are the intensities of two different primary colors.
46. The apparatus of claim 45, wherein a is red color and b is blue color.
47. The apparatus of claim 45, wherein a is blue color and b is green color.
48. The apparatus of claim 45, wherein a is green color and b is red color.