Device for Optically Reading a Symbol on a Form by Shining Radiation Therethrough, and Related Method

A vast number of documents are issued every day on pre-printed forms. Those forms often have identification numbers or symbols printed on them so that when information is written onto the form, that particular form can he identified and the information recorded on that form stored and retrieved by reference to its identification number. Common examples are personal checks. The blank forms may he identical except for the check numbers. The check register is organized by check number, and the date, amount and payee are recorded according to check number. Likewise a money order may he written onto a pre-printed form with an identification symbol on it which represents the identification number of the form. As more and more of these documents, including personal checks, are being written onto pre-printed forms by computer printers, there is a great need for an inexpensive and reliable system for machine reading an identification symbol printed on those forms.

If the identification numbers of those forms are manually recorded at the time the forms are filled out, the process is cumbersome, time consuming, expensive and prone to inaccuracy due to human error. Therefore a system of machine reading and entry of the form identification is preferable.

One system of machine reading an identification symbol is the bar code system which uses a machine to read a bar code printed on a document. Such systems are often designed for use in identifying documents or packages, such as mail, where the reader must be able to read the bar code independent of the relative position of the bar code and the reader. While satisfactory for their intended purposes, for some applications, such bar code reading systems are less than ideal because the reader generally requires a lens to focus a reflected light beam onto the bar code, onto the sensor, or onto both, often requires the use of a coherent light source, and generally requires movement between the light source and the code, usually generated by a rapidly moving part such as a spinning mirror or spinning hologram. Each of these requirements adds to the expense of manufacture and leads to errors during the use of such a reader. In addition to these shortcomings with the reader, the symbol used in such a system generally must be printed to precise specifications and therefore may be relatively expensive to print and subject to error.

Many systems exist in the printing of information on forms to generate documents where the forms are similar to each other and their position and movement relative to a mechanical device, for example a tractor-type printer, is predictable. Mounting a sensor on such a device would eliminate the need for a system such as a bar code system that allows "reading" the code regardless of the relative position of reader and the document, thus avoiding the cost and inaccuracy of such a system.

This invention is directed to an inexpensive and highly reliable apparatus and method for machine reading symbols printed on forms. It comprises a device for machine reading a symbol on a document which document is at least partially translucent to incident radiation and which has markings on it in the form of a symbol. The markings are transparent to the incident radiation to a different degree than the unmarked portion of the form. The reader has a radiation source such as a light source which shines radiation through the document at the location of the symbol and onto a sensor such as an array of light sensitive cells. The array generates a signal in response to the radiation falling on it.

The shape of a symbol printed on the form varies depending on the information (such as an identification number) encoded in that symbol. The pattern of light and dark created when light shines through that symbol therefore varies, and the signal generated by the light falling on the array varies in accordance with the shape of the symbol. Therefore that signal may be processed to decode any information encoded in the symbol.

The symbol may be a combination of several patterns, for example, patterns comprising parallel areas of pigmented bars and unpigmented spaces. Each pattern may represent a digit. When several such patterns are aligned end-to-end directly abutting each other, they may create, for example, a symbol that represents a multi-digit number.

The form is advanced, smoothly or in a step-wise fashion, between the sensor and the light source. As the symbol printed on the form is passed between the sensor and the light source, the signal from the sensor is periodically sampled, and the signal so sampled is processed to read each pattern. The signal representing each pattern might be sampled several times, and each such signal processed to decode the pattern.

Each pattern is decoded to determine its value, e.g. a digit, and the value of the patterns that make up the symbol are combined to determine the information encoded in the symbol, e.g. a multi-digit identification number. If the signal representing each pattern is read several times, only the most reliable reading is used to combine with other readings to determine the information stored in the symbol. That information may then be recorded in conjunction with the information printed on the form for later retrieval.

The method of the invention comprises the use of the document reader described above to read symbols printed on the form.

The reader of the invention and its use thus allows for an inexpensive machine reader to read symbols printed on forms.

The reader of the invention and its use provides for a highly reliable means of machine reading symbols printed on forms.

The reader of the invention and its use provides for machine reading a symbol on a form while the form is being advanced through a printer.

The reader of the invention and its use provides for machine readings of a pattern while the pattern is stationary relative to the reader.

The reader of the invention and its use provides for combining multiple patterns to store and read information in a symbol.

The reader of the invention and its use provides for a machine reader which may be lens-free.

The reader of the invention and its use provides for the generation of an identification signal that may be processed in a manner which produces a highly reliable result.

These and other objects and advantages of the invention and the manner of achieving them will be further explained and may be further understood in conjunction with the following illustrations and detailed description.

FIG. 1 is a perspective top view of an embodiment of the invention including an optical reader and a form having a symbol printed thereon, the form shown partially cut-away;

FIG. 1A is an elevated view of the linear array of light sensitive cells;

FIG. 2 is a top view of a form for use in the invention;

FIG. 3 is a perspective top view of the machine reader and forms of the invention mounted in a tractor printer;

FIG. 4 is a diagrammatic representation of the invention with flow charts indicating the signal processing of a preferred embodiment;

FIG. 5 shows the patterns used in a preferred embodiment;

FIG. 6 is a representation of the patterns shown in FIG. 5 combined to form an identification symbol.

As is shown in the drawings for purposes of illustration, the embodiment depicted uses an optical image detector (11) incorporated into a tractor-type printer mechanism (13) for use with forms (15) having an identification symbol (17) printed on each form. The form is advanced along the paper path of the printer and information is printed onto it by the print head (19). For example, if the form is a money order, the date, amount and payee may be printed onto the previously blank form. As the form is advanced along the paper path, the identification symbol printed on it is passed through the optical image detector, a signal is generated. That signal may then be processed to read and decode the information encoded into the symbol, and that information may then be recorded in conjunction with the other information printed on the form.

The Optical Sensor

The optical image detector (11) comprises a light source (21) and a light sensing array (23) spaced apart a sufficient distance for a form to be passed between them. The light source may be, for example, a light emitting diode. The light emitting diode is mounted in a housing (25) which holds it near the paper path, for example, just below the lower surface of the paper path. It holds the LED so that the LED directs light (27) toward the light sensing array. The housing for the LED may also incorporate a spring (28) for urging the form into close proximity with the light sensing array. The LED generates light (29) of sufficient intensity so that when it is transmitted through the form, it is sufficiently intense to be detected by the light sensing array.

The light sensing array (23) is mounted in a housing (31) which holds the array near the paper path in the tractor-type printer directly opposite the light source, for example, just above the upper surface of the form (15) as it advances into the tractor-type printer. The sensor may be, for example, a charge coupled device ("CCD") having 64 light sensitive cells (32) in a single file linear array, each cell being 1/8 mm square. A reference voltage (33), for example 5 volts, is applied to this CCD. The CCD array may be contained in a an electronic chip which also has self-contained clocking circuitry which may be utilized in determining the period of sampling of the signal generated by the CCD. A commercially available example of a chip of the type described here is the TSL 214 manufactured by Texas Instruments.

Each cell within the array generates a signal whose intensity varies in relation to the amount of light falling on the cell. The signal of each cell may be combined with the others from the array to form an analog signal (37) representing the pattern of light intensity falling on the entire array. Therefore, as the form is passed between the LED and the CCD, a signal is generated that represents the pattern of light intensity transmitted through the form. The signal is sampled periodically at periods determined by the CCD clock generator (39), which as mentioned above, may be contained in the chip incorporated into the CCD. The signal sampled will represent the pattern (41) interposed between the LED and CCD at the time of the sampling, and may be processed to decode any information stored in that pattern.

The form (15) used in the machine reader of this embodiment is a pre-printed paper form such as a blank money order (43). The substance that is used for the form may be safety paper or the like to deter forgery, plain paper, or any substance, provided whatever is used is sufficiently translucent to light (29) from the LED (27) that sufficient light is transmitted therethrough to generate a signal when falling upon the cells (32) of the CCD array. The indita printed on the form, other than the symbol (17) is not critical, except that it may not obstruct light passing through the symbol. This may be accomplished, for example, by having a zone (45) on the form including an area (47) wherein the symbol is located, and having the zone being slightly larger than the symbol, the zone being entirely blank except for the symbol.

Each form has a symbol (17) printed on it. The symbol in this embodiment is comprised of parallel pigmented areas (49) and unpigmented areas (51). The pigmented areas of the symbol are less transparent to light than are the unpigmented areas so that when light shines through the form (15) at the location of the symbol and onto the light sensing array (23), the symbol acts essentially as a mask so that a sharp image of light and shadow is cast onto the array even without a lens to focus the image. The intensity of the light falling on each cell of the array varies depending on whether the portion of the symbol between the cell and the (21) is pigmented or not. In turn, the cells generate a signal that varies depending on the intensity of light falling on each cell. When all the signals of the cells are combined, the resultant aggregate signal may be processed to determine the information encoded in that portion of the symbol.

The Symbol

The symbol (17) printed on the form may be any appropriate marking which may act as a mask to create an image of light and shadow on the sensor when the symbol is moved between the light source and the sensor. The symbol described here which has been found useful is made by combining a plurality of patterns (53) oriented end to end, directly abutting one another in vertical alignment. In this embodiment, there are 11 unique patterns, with one representing a start code (55) and one each representing the numbers zero (57) through nine (59). These patterns are in the form of a series of parallel stripes which are either bars (pigmented stripes) (49) or spaces (unpigmented stripes) (51). When a plurality of these patterns are oriented in end-to-end abutting juxtaposition, the resultant symbol represents a sequence of digits which together encode a multi-digit number (see generally FIG. 6). While always the same in any particular application, for different applications, the precise size limitation of the bars and spaces is not critical to the invention except that it must create a variable pattern of light and dark that carries any information within the width that falls on the light sensor. In the embodiment depicted here, the bars are slightly different in width than the spaces for ease of printing. The width dimensions of the bars and spaces are set out in the table below:

The patterns which are used in this embodiment which represent a start code and the digits zero through nine are all shown in FIG. 5.

The length of individual patterns (53) is determined in a manner appropriate for the paper movement and signal sampling employed. For example, the patterns depicted are used in a tractor-type printer (61) that advances the paper in steps of 0.110 inches. The signal generated by the sensor is sampled after each advance of the tractor-feed mechanism. The patterns are each 0.219 inches long. By selecting a pattern length of 0.219 inches, the signal may be sampled in a manner to assure that the signal selected does not represent a portion of the pattern which is near the border (63) between two adjacent patterns. This is desirable since it enhances the reliability of the sensor reading. This is explained in greater detail in the next section.

The Signal Processing

As an overview, and with reference to FIGS. 3 and 4, the signal (37) generated by the array (23) is sampled periodically at times dictated by the CCD clock generator (39), and the signal so sampled is processed by the signal processor. The signal generated by the array is an analog signal. That signal is converted by an analog/digital converter (67) into a digital signal. That digital signal is then examined by the image processor (69) to determine whether it represents a start code (55). If no start code is detected, no action is taken. If a start code is detected, then a certain number of the signals subsequently sampled are processed further. For example, if the number of digits in a multi-digit number is known, and a start code is detected, then the subsequent readings are processed to determine each subsequent digit until all the digits in the number are determined; if the form contains a ten-digit identification number, then ten digits subsequent to the start code are determined.

The identification of the image of light and dark, and the assignment of a value to that image is performed by the image processor (69). The precise method of determining the value of each digit is done according to the computer program which is set out as an Appendix, submitted herewith.

High reliability is a valued feature, and several methods of assuring high reliability are incorporated into the signal processing of this embodiment. One such method is to read each pattern twice, decide which reading is most likely to be reliable, and to select that reading to be used to determine the digit represented by the pattern.

One method of doing this is by analyzing the first reading after a start code (55) is detected. That next reading will either be another start code, a nonsense or ambiguous reading (generally indicating that the sensor is positioned over a boundary between patterns (63)), or an identifiable digit code. The proper readings to be relied upon thereafter are determined as described here.

The signal (37) generated by the optical sensor array (23) is most likely to be ambiguous or nonsensical if the optical sensor is reading the symbol at a boundary between two patterns (63). In order to minimize this likelihood of reliance on such a signal, a method of selecting a preferred signal is used based upon the length of each advancement of the printer. The tractor-feed printer advances the paper by a length of 0.110 inches each step. The start code is 0.163 inches long, slightly more than 1 paper advance in length; each pattern representing a digit is 0.219 inches long, slightly less than 2 paper advances in length. After the signal processor first detects a start code, it samples and processes the signal after each subsequent tractor-feed advance, i.e., after the form is advanced 0.110 inches. If the second reading is also a start code, then that reading is 0.110 inches from the initial reading and thus near the boundary between the start code and the first pattern representing a digit. The next reading (the third reading) will be 0.110 inches farther along and thus not near the boundary. This reading, therefore, is the preferred reading in determining the digit encoded in that first pattern since it is deep into the pattern and not near a border between patterns.

A reading after the next advance (the fourth reading) will be at a place close to the border between the first and second patterns since each pattern is 0.219 inches long. It is therefore not the preferred reading for the second pattern. The following reading will be preferrable. Thus, in this example, the odd numbered readings (first, third, fifth, etc.) beginning with the first reading that is a digit, are the preferred readings.

If the first reading is a start code, and the second reading is a nonsense or ambiguous reading, the second reading is likely on the border between the patterns for the start code and the first digit. The paper advance of 0.110 inches will position the reader deeply into the first pattern. Each successive two advances will likewise position the reader over a pattern for a digit deep within (i.e., far from the border of) a pattern. In this case, the more reliable readings will be the even numbered readings after the start code is first detected.

If the first reading after the start code is an identifiable digit, then the reader at the time of that second reading is positioned somewhere within the first half of the pattern but deeply enough into the pattern to generate a readable signal. Two additional advances will position the reader even more deeply Within the next pattern since two advances move the symbol 0.220 inches and the pattern is only 0.219 inches long. In this instance, as in the first example, the preferred readings will be the odd numbered readings beginning with the first reading after the start code that generates a signal that is "read" as an identifiable digit.

The preferred readings of each digit (e.g., readings 1, 3, 5, etc.) are stored. The values determined for the alternate readings (e.g., readings 2, 4, 6, etc.) are stored separately. The values of each of the preferred readings are then combined to form a multi-digit number (e.g., 1, 3, 5, . . . ). If, for any reason, the value of one of the preferred readings is nonsensical, then the image processor substitutes the alternative reading of that pattern (e.g., if the preferred reading for the third digit is nonsensical, the alternative reading for the third digit is substituted). Likewise, if any of the alternative readings are nonsensical, the preferred reading for that digit may be used.

Once a preferred list of digits and an alternative list of digits is determined, a check sum is used. The check sum is employed as follows. The last pattern in a symbol (71) represents a digit that is not itself part of the multi-digit identification number. Rather, that pattern is selected to represent a digit which, when added to all the other digits represented in the symbol, results in a sum that is evenly divisible by nine. Therefore, upon completion of reading the last digit in the symbol, the image processor sums all the digits of the preferred list and divides the sum by nine. If the sum is evenly divisible by nine, the multi-digit identification number is assumed to be correct and the determination is complete. That number is used (e.g., stored as the form identification number).

If the sum is not evenly divisible by nine, then the image processor will sum all the digits in the alternative list. If that sum is evenly divisible by nine, then that list of digits is assumed to be correct, and is used to form the multi-digit identification number. If not, one final attempt is made to determine the value of that identification number.

The image processor has an expected identification number for each form based on the previously determined number. If neither of the check sums are evenly divisible by nine, the image processor examines the preferred value for the last three patterns (including the check sum pattern). If any two of these digits are the digits expected by incrementing the previously determined value by one, then the image processor assigns as the form identification number the expected number (usually the previously identified number plus one). If not, the image processor may signal the operator in some appropriate manner.

This check sum procedure results in a determination in a high percentage of attempts and with a high degree of reliability.

In operation, forms (15) with an identification symbol (17) on them are loaded into a printer (61) which incorporates an optical image detector (11). (If they are in sequence, the initial form number is input into a micro-controller (65) such as a computer. This number is incremented by one each time a form is advanced. The micro-controller thus has an expected identification number for each form.) Each form, as needed, is advanced through the printer in step-wise fashion by a tractor-feed mechanism and passes between the light source (21) and the light sensing array (23). As the form is advanced, the symbol (17) moves between the light source and the array in close proximity to the array, creating an image of light and shadow that represents the pattern printed on the form. The array generates a signal in response to the light falling on it. The signal is sampled after each step-wise paper advance, and the signal is decoded by the image processor (69) to determine the value of the digit encoded in the pattern (53). The number of digits appropriate for the multi-digit number are combined to determine that number as encoded in the symbol. The number so determined is then stored in conjunction with the information subsequently printed on that form for later retrieval and use. For example, if the form is a money order, the payee, amount, date, and location of the issuance, and the like, can be stored and later retrieved by reference to the identification number encoded in a symbol printed on the form.

While specific embodiments of the invention have been described in detail, it will be appreciated by those skilled in the art that various modifications to the structure and use of the disclosed invention may be made in light of the overall teachings of the disclosure, without departing from the scope of the invention as defined by the appended claims. ##SPC1##