Bar Code and Method and Apparatus for Interpreting the Same

1. Field of the Invention

The present invention relates to information bearing bar codes, and more particularly, to bar codes of the type which are arranged to be optically scanned by means of a hand-held light pen or wand. It is a primary object of the invention to provide a new and improved bar code arrangement wherein individual characters of the bar code are extremely small and may be detected by means of a decoding arrangement which is substantially independent of the manual scanning speed with which the light pen is moved over the label or ticket containing the bar code, while at the same time providing an arrangement wherein the code may be accurately scanned without repeated passes over the label and with a minimum of substitution and transposition errors.

2. Brief Description of the Prior Art

It is well known to use an optical scanner to capture data from a bar code. Typically, the optical scanner includes a source of illumination and a photodetector. The scanner is manually drawn across a bar code comprising a series of variable-width bars and spaces. During the scan, apparatus is provided to measure the time which it takes the optical scanner to pass over each bar and space within the code. The resulting series of time measurements correspond roughly to the width of the bar and space elements of the code and may be decoded to represent the data content of the code. Such a system is disclosed in U.S. Reissue Pat. No. Re. 28,198 which issued to the present inventor on Oct. 15, 1974.

Systems of the type just described have come into relatively widespread use. They are used, for example, to bar code price tags at the checkout counters of department stores and also to read inventory labels in automated industrial inventory-control systems. Customarily, the bar code is printed in association with a human readable set of characters so that the purchaser can himself identify the price, and other information associated with the article bearing the label.

While these systems are satisfactory for their intended purposes, the individual who handles the scanning stylus must be careful to draw the tip of the stylus directly over the printed bar code from one end of the code to the other. If the stylus slips off the edge of the bar code, then the data is not read correctly. Similarly, if dirt spots, gouges, or other irregularities are scanned, data is not normally collected and the scan must be repeated. Accordingly, it is sometimes necessary to scan the same bar code a number of times before the encoded data is properly read.

Also, with conventional bar code readers it is also necessary to have the length of the bars several times the height of the accompanying characters which are human readable so that the bar code occupies a relatively large space and requires a relatively large label.

Since the velocity with which the tip of the stylus moves over the printed bar code varies with different operators, the time taken to pass over each bar and space may vary over quite wide limits. In order to correct for such velocity changes, it is customary to store all of the digits of a particular character and from such storage determine through suitable logic the actual width of the bars and spaces. While various arrangements have been proposed for such velocity correction, all of them require additional logic and the storage of a large number of bits of information, all of which increases the complexity of the decoding apparatus. It would, therefore, be desirable to provide a scanning arrangement wherein the decoding of the bar code is independent of the velocity with which the manual scanner is drawn over the bar code.

Various other arrangements have also been heretofore proposed for scanning labels at constant velocity wherein a moving label is scanned by a rotating mirror, or the like. Such arrangements are shown, for example, in Stites et al U.S. Pat. No. 3,225,177 and Wolff U.S. Pat. No. 3,744,026, but are unsuitable for adaptation to hand-held scanning units due to the size, weight and complexity of such arrangements.

In addition to the above described bar code readers, it has also been proposed to provide only human readable characters on the label which are read directly by means of a hand held scanning unit. These arrangements have employed a photoelectric cell matrix on which is optically focused each OCR character as the scanner unit is moved across the label. Each character is compared with a series of masks corresponding to the characters to be recognized and when the scanned data corresponds to a particular mask, entry of the corresponding character is made. Such an OCR type of hand held scanning arrangement requires a relatively complex decoding arrangement for scanning the matrix in both horizontal and vertical directions and in order to provide a system which is not too expensive, the number of characters which can be recognized is limited to approximately twenty-one, so as to simplify the character recognition decoding process. The use of only twenty-one characters severely limits the clarity with which a message may be encoded by this method since a large proportion of characters are required for numbers, thus leaving only a few letters to convey the remaining desired information in human readable terms. Furthermore, the degree of skew which the hand scanner may assume with respect to the scanning path is severely limited in such OCR readers and a maximum skew limit of eight degrees may, in some instances, be required. Also, since the photoelectric cell senses the label over a substantial area and hence over a substantial width along the scanning path, the hand-held scanner unit must be maintained perpendicular to the plane of the label as the scanning operation is performed since otherwise portions of the light pattern striking the matrix will become defocused and reduce the reliability of detection of the character. As a result, the geometry and accuracy of printing which is to be read by such OCR hand scanners must be held to a very close tolerance. In addition, there is no way for the OCR reader to detect that an improper substitution has been made in the decoding process. Thus, if due to voids or smears in the printing operation a "2" becomes damaged so that it is read as a "3", there is no detection of this error and the wrong price may be decoded. Furthermore, since these voids and smears may be extremely small the operator who is using the hand scanner cannot visually detect the damage and hence is unaware of the substitution error.

Systems of the type just described have come into relatively widespread use. They are used, for example, to bar code price tags at the checkout counters of department stores and also to read inventory labels in automated industrial inventory-control systems. These systems are highly satisfactory. However, the individual who handles the scanning stylus must be careful to draw the tip of the stylus directly over the printed bar code from one end of the code to the other. If the stylus slips off the edge of the bar code, then the data is not read correctly. Similarly, if dirt spots, gouges, or other irregularities are scanned, data is not normally collected and the scan must be repeated. It is sometimes necessary to scan the same bar code a number of times before the encoded data is properly read.

It is, therefore, an object of the present invention to provide a new and improved bar code system which eliminates one or more of the data discussed disadvantages of prior art arrangements.

It is another object of the present invention to provide a new and improved bar code system that is less sensitive to the positioning of the stylus relative to the code than prior arrangements.

It is still another object of the present invention to provide a new and improved bar code system in which a hand scanner unit is manually moved across the different bar code characters and decoding is substantially independent of the velocity of such manual scanning.

It is also an object of the invention to provide a scanning apparatus which can capture all the data encoded into a bar code in a single pass of the stylus. Additional objects of the invention are to provide a bar code and scanning system that are considerably less sensitive to the occurrence of minor defects in the bar code than are conventional systems and that are also more insensitive to variations in the speed at which the bar code is scanned by the stylus -- particularly to sudden changes in the speed of scanning. Desirably, the scanning system should be capable of capturing the encoded data regardless of how fast or slow the scan may be, unlike conventional arrangements which require a swift, relatively steady scanning motion that is not too fast.

Briefly described, the present invention contemplates providing a system in which the individual encoded characters are each scanned repeatedly by a light-sensing linear photodetector array which is arranged to scan in a direction perpendicular to the direction in which the hand-held stylus is moved. Preferably, the photodetector array is arranged to scan repeatedly across regions containing successive bars and spaces while the manual scanning motion moves the photodetector array from one character to the next. This system has the significant advantage over prior bar code hand scanning arrangements that the speed of scanning in a direction perpendicular to the bars and spaces is determined by the speed at which the linear photodetector array is electronically scanned and not by the speed of the manual motion of the hand-held stylus.

In accordance with the preferred embodiment of the invention, each individual character which is to be encoded is represented by a series of width-modulated bars and spaces which are vertically stacked with the bars and spaces all horizontal. The bars and spaces used to represent different characters within a message are displaced from one another horizontally and a vertically extending bar which joins all of the horizontal bars at one edge thereof is employed to facilitate separation of characters. A hand-held stylus is then moved horizontally across the characters and contains a light source, a linear photodetector array and suitable optics for focusing light reflected from a single vertical line in the label or ticket so that all elemental areas along this line simultaneously energize the linear array of photodetectors. An electronic scanning arrangement is employed to develop a pulse type output in which each pulse corresponds to one of the photodetectors in the linear array and the amplitude of each pulse is proportional to the amount of light striking that photodetector. Accordingly, the speed of the stylus horizontal motion has vertually no effect upon the interpretation of the code, since the relative positioning of the photodetectors along said vertical line is fixed and the pulses are generated at fixed intervals dependent only on the electronic scanning rate and not on the manual horizontal motion of the stylus.

The linear photodetector array may be a charge-coupled device of conventional design or any equivalent device. Preferably, the array should be self-scanning, as is conventional in such devices, so that the array generates an amplitude modulated pulse type signal representing the pattern of light and dark patches along the vertical line which is scanned. Preferably, the vertically stacked bars and spaces occupy relatively little space vertically and have a height approximately one-third to one-fourth that of conventional bar codes. Also, the length of these vertically stacked bars may be quite small since the rate at which the linear array of photodetectors are scanned electronically can be sufficiently high that each character is scanned many times at normal velocity of the hand-held scanner even with relatively short bars. Accordingly, the total area for each character is much smaller than that of previous hand scanned bar codes and each character may occupy a space equivalent to an OCR character so that extremely small labels may be produced in accordance with the present invention.

The output signal generated by the array is processed to eliminate signal fluctuations which are characteristic of small spots and voids by means of a special digital filter. The resultant digitized and filtered signal may then be analyzed to detect the presence of bar and space combinations which correspond to a valid character. The system is preferably set up so that a series of consecutive scans must always result in the detection of the same character before the scans are accepted as a valid reading of an encoded character. In this way, a high degree of assurance is given that the data captured is correct. The use of a spot and void filter in conjunction with the scanning of each character at many different points insures that in almost every case a valid reading of each character is assured.

Further objects and advantages of the invention are apparent in the detailed description which follows. The points of novelty which characterize the invention are pointed out with particularity in the claims appended to and forming a part of this specification.

For a better understanding of the invention, reference will be made to the drawings wherein:

FIG. 1 is a perspective view of a scanning stylus designed in accordance with the preferred embodiment of the invention and suitable for manual scanning -- the stylus is shown scanning a bar code that is also in accordance with the preferred embodiment of the invention;

FIG. 2 is a front elevational view of the stylus shown in FIG. 1;

FIG. 3 is a schematic representation of the stylus shown in FIG. 1 illustrating its internal constructional details;

FIG. 4 is a block diagram of the preferred embodiment of the present invention;

FIG. 5 is a logic diagram of the spot and void filter shown as a block element of FIG. 4;

FIG. 6 is a logic diagram of the data buffers and error-check logic which appear together as a single block element of FIG. 4;

FIG. 7 is a partly block and partly logic diagram of the character range logic which appears as a block element of FIG. 4;

FIG. 8 is an illustration of a typical encoded message using a bar code in accordance with the preferred embodiment of the present invention and including a variety of voids, ink spots and other irregularities;

FIG. 9 illustrates the elements of a linear photodetector array upon which have been simultaneously projected a bar and the adjoining spaces to illustrate how the spot and void filter shown in FIG. 5 may be used to cancel out the signals which result from the scanning of small spots or voids within a printed bar code;

FIG. 10 illustrates an array of photodetectors upon which have been simultaneously projected an image of a series of three bars and the adjoining spaces; and

FIG. 11 illustrates the output signal which is developed when the series of photodetectors shown in FIG. 10 are scanned.

With reference to FIG. 8, the preferred embodiment of the present invention contemplates providing a bar code in which the bars and spaces used to represent each character are oriented vertically, as shown, and in which the bar and space combinations used to represent different characters are spaced from one another horizontally. In FIG. 8, for example, the letter A is represented by a series of bars 802 and spaces 803 which comprise a bar-encoded character 801. The bars 802 and spaces 803, whose relative widths represent the letter A, are stacked vertically -- one on top of another -- such that one wishing to decode the information represented by the bars and spaces must scan them in a vertical direction, either upwards or downwards. The letter "A" itself is shown at 802 printed directly below the bar-encoded character 801. It is noteworthy that the bar-encoded character 801 is just as compact as the printed letter "A" at 802. In prior art arrangements, it has been customary to utilize vertical bars and spaces and to have the length of the bars and spaces be four to five times the length of the corresponding printed letter or numeral to make it easier for the bars and spaces to be scanned by means of a hand-held stylus. The present invention obviates the need to have the bars and spaces extended to such lengths.

At the righthand edge of the horizontal bars 802 and spaces 803, a single vertical bar 804 is provided which connects to each of the bars 802 to give the resultant bar-encoded character 801 an appearance which roughly resembles a backwards letter "E" containing an extra horizontal stem. A vertical space 805 is left between the adjacent characters 801 and 806. Directly beneath each of the bar encoded characters 801, 806, etc. are the corresponding human readable characters 802, 807, etc. which are laid out as shown so that they may be easily read. The bar-encoded and human readable characters may be printed or typed simultaneously, if desired, and the spacing between adjacent characters is not critical.

In accordance with one aspect of the present invention, it is contemplated that a stylus 101 (FIG. 1) is to be provided which may be drawn horizontally by hand across bar-encoded characters of the type shown in FIG. 8. Within the scanner 101 there is mounted a linear array of photodetectors which are shown at 301 in FIG. 3. This array is of a conventional type. For example, it may be the Model CCD101 solid state linear image sensor 500-element charge-coupled device which is presently obtainable from the Fairchild Semiconductor Components Group of the Fairchild Camera and Instrument Corporation, 464 Ellis Street, Mountain View, Calif. This device includes an array of 500 photodetectors arranged in a row to sense light which is projected thereon. An electrical charge is developed within each of these photodetectors proportional to the quantity of incident light received by each photodetector. At periodic intervals, this charge is shifted laterally into an analog charge-transport shift register and is conveyed in bucket-brigade fashion to the edge of the array where it is amplified and presented as a series of amplitude modulated pulses to an output signal line. The shift register operation is controlled by pulses which come from an external clock.

Within the hand scanner 101, there may be mounted an array of light sources, for example, light emitting diodes, which are indicated generally as 302 in FIG. 3. These diodes illuminate a narrow vertical strip 303 immediately beneath the stylus 101. An image of the vertical strip 303 is reflected by a mirror 304 to a lens 306 which focuses the image of the strip 303 upon the array of photodetectors 301. FIG. 10 illustrates schematically the array of photodetectors 1001 upon which has been projected an image of the bar-encoded character 801 shown in FIGS. 1, 3 and 8. The projected image of the bars 820 results in dark regions 802' which cover some of the photodetectors 1001, while the projected image of the spaces 803 results in bright regions 803' which cover the remaining photodetectors 1001. FIG. 10 illustrates that some of the photodetectors receive light while other do not receive light, in accordance with the precise positioning of the bar code character 801 with respect to the position of the stylus 101.

When the individual photodetectors 1001 are scanned, the resulting array output signal appears as is shown in FIG. 11. A comparison of FIGS. 10 and 11 reveals that this signal accurately depicts the light and dark regions of the bar encoded character 801. Each vertical spike in the signal shown in FIG. 11 represents the charge collected by a single photodetector, and the height of each spike indicates how much light the corresponding photodetector received during the immediately preceding period. Much higher amplitude spikes result from photodetectors upon which the space images 803' are projected than from photodetectors upon which the bar images 802' are projected. While the amplitude levels of the signals shown in FIG. 11 are preserved, the signal is somewhat quantized from left to right in that each of the spikes shown in FIG. 11 corresponds to the output of a single photodetector shown in FIG. 10. The Fairchild CCD101 image sensor separates these spikes in synchronism with the receipt of pulses from an external clock. Hence, it is possible, if one knows the geometry of the stylus 101 and the number of pulses per second supplied by the external clock, to determine precisely how wide the bars 802 and spaces 803 are by measuring the time durations of the high and low signal regions within FIG. 11. It is therefore, possible by simply timing the duration of the high and low pulse regions of the waveform shown in FIG. 11 to get a precise measurement of the absolute widths of the successive bar and space elements of each character. With a conventional hand-held stylus that contains just a single photodetector, it is in most cases impossible to measure the absolute width of bars and spaces, and only relative width measurements may be made. This requires storage of the measurements of all of the bars and spaces in a character, as described in U.S. Reissue Pat. No. Re. 28,198 so that relatively complex decoding logic is required as compared to the present invention.

FIG. 4 is a block diagram overview of a scanning system 400 designed in accordance with the present invention. The system 400 includes a system clock and array scan logic 401 which is used to control the operation of the linear array of photodetectors 301 mounted within the scanner 101. The details of the system clock and array scan logic 401 are conventional -- it simply generates those signals which are necessary to cause the linear array of photodetectors 301 to function and to generate an output waveform such as that shown in FIG. 11 in synchronism with the operation of the system clock at 401. Schematic and logic diagrams of the required logic may be obtained, for example, from the Fairchild Semiconductor Components Group whose address is given above. The Fairchild schematic and logic diagrams would be suitable for use with the Model CCD101 image sensor device which Fairchild currently markets. If a different type of image sensor array is used, the constructional details of the necessary clock and scanning logic similarly may be obtained from the manufacturer of the linear array.

As the system clock 401 advances, it supplies pulses to an array address counter 402. The array address counter 402 generates a binary number which indicates from which photodetector within the linear array of photodetectors 301 data is being retrieved at any given moment and supplied as an output signal of the linear array. When the array address counter 402 reaches full count, it generates a reset signal which resets the system clock and erase scan logic 401 and may also initiate a new scan of the array 301. Alternatively, the system clock and array scan logic 401 may reset the array address counter after each scan is completed.

The output signal of the linear array 301 is first fed through a conventional analog-signal low-pass filter which smooths over the spikes that appear in FIG. 11. The smoothed-over signal is then fed to an analog-to-digital converter 403. The purpose of the converter 403 is to convert the variable-amplitude filtered analog signal into a two-state signal that is always either high or low and that makes clean transitions from its high state to its low state and vice versa. Such converters are well known to those skilled in the art. A typical such converter, for example, is disclosed in U.S. Pat. No. 3,849,632 which issued on Nov. 19, 1974.

The output of the analog-to-digital converter 403 is thus a signal which closely resembles the signal shown in FIG. 11 except that the higher spikes have been replaced by a continuous high-level signal and the lower spikes have been replaced by a continuous low-level signal, the high and low fluctuations of the resulting two-state signal being clean and well defined.

This two-state signal is fed into a spot and void filter 500, the details of which are shown in FIG. 5. The spot and void filter 500 is driven by a CLK signal from the system clock at 401. In the preferred embodiment of the invention, the spot and void filter 500 comprises a series of three type D flip-flops 502, 504 and 506 which are interconnected to form a three-bit shift register, as is shown. The three outputs from these flip-flops on one side are fed into an AND-gate 508 the output of which feeds the J input of a JK flip-flop 510. The three remaining outputs of the three flip-flops are fed into an AND-gate 512 which feeds the K input of the JK flip-flop 510. The system clock drives all four of the flip-flops in synchronism.

The purpose of the spot and void filter 500 is to refuse to recognize any fluctuation of the output signal delivered by the analog-to-digital converter 403 unless that fluctuation persists for at least three system clock timing intervals after it first occurs. Assuming that a high-level signal has been present at the input to the flip-flop 502 for some time, all of the flip-flops 502, 504, 506 and 510 will be in their "set" states. Assume now that the input signal to the flip-flop 502 from the analog-to-digital converter 403 now goes to a low level. The next incoming clock pulse clears the flip-flop 502 and causes it to supply a low-level signal to the AND-gate 508 and a high level signal to the AND-gate 512. The signal supplied to the AND-gate 512 is not permitted to reach the flip-flop 510 because the two flip-flops 504 and 506 continue to disable the AND-gate 512. Hence, this fluctuation of the input signal to the flip-flop 502 does not toggle the flip-flop 510 immediately. The next succeeding clock pulse clears the flip-flop 504 so that it, too, supplies a high-level signal to the AND-gate 512. However, the flip-flop 506 continues to disable the AND-gate 512. A third clock pulse finally clears the third flip-flop 506. Now the AND-gate 512 is fully enabled to supply a high-level signal to the K input of the flip-flop 510 and the next-to-occur clock pulse clears the flip-flop 510. The net result of all this is a brief time delay, but the output signal which flows from the flip-flop 510 is identical to the input signal to the flip-flop 502 in all other respects. If the input signal to the flip-flop 502 had not stayed at a low level for at least three clock pulses, the signal would have gone to a high level and permitted the flip-flop 502 to be set before the flip-flop 510 was set. The flip-flop 502 would then have disabled the AND-gate 512 and prevented the toggling of the flip-flop 510 into its cleared state. Hence a short-term low-level fluctuation of the input signal to the flip-flop 502 would have been ignored and would not have reached the output of the flip-flop 510.

The significance of the digital-filtering arrangement just described lies in its ability to cancel out the effects of small voids and ink spots, as is apparent in FIG. 9. The array of photodetectors 1001 is therein shown as having projected upon itself a bar image 900 and images of the two adjoining spaces. The bar image 900 includes the image of a small void at 901 which is large enough to cover three of the photodetectors 1001. An ink spot at 902 within an adjoining space region is also large enough to cover three of the photodetectors 1001. When the array of photodetectors 1001 is scanned, signal fluctuations at the inputs of the spot and void filter will occur corresponding to the void 901 and the ink spot 902 which endure for only three clock pulses -- the time required to scan three photodetectors -- and which, therefore, are not able to toggle the flip-flop 510. Such tiny flaws within a bar-encoded character, which may not be visible to the human operator, thus cause no fluctuations of the signal which is ultimately supplied to the transition detection logic 404 (FIG. 4) and thus cannot cause an erroneous reading of the encoded data. The spot and void filter is able to distinguish a spot or void from a narrow bar or space only because the scanning carried out by the linear array is able to measure the absolute width of the spots and voids and distinguish spots and voids from bars and spaces by the narrower width of the former. Systems which measure only the relative width of spots and voids and of bars and spaces cannot distinguish spots and voids from narrow bars and spaces.

The output signal of the spot and void filter 500 is fed directly to the transition detection logic 404. Briefly described, the transition detection logic 404 senses fluctuations of the incoming signal and responds to such fluctuations by supplying pulses to a character separation logic 405 which includes a mechanism for measuring the time that elapses between the successive pulses and therefore between successive bar-space and space-bar transitions. The character separation logic 405 preferably compares the relative widths of the bars to each other and also compares the relative widths of the spaces to each other. The character separation logic generates digital data which indicates the results of these width comparisons (wider than, narrower than, or the same size as) and supplies this digital data to a decoding logic 406 which decodes the relative width data into the actual data content of the encoded character. The decoding logic 406 also rejects all bar-code relative width data that does not correspond to the scanning of a valid character and thus prevents, for example, a scan of the human-readable character "A" in FIG. 8 from being interpreted as a bar code. The elements 404, 405 and 406 of the present invention are conventional width-modulated bar-code decoding logic elements examples of which may be found both in U.S. Reissue Pat. No. Re. 28,198 and also in a patent application Ser. No. 399,773 filed by Arnis G. Butulis on Sept. 20, 1973 and assigned to the same assignee as the present application. Within the Butulis application, the transition-detection logic 404 is shown at the far left-hand edge of FIG. 2 wherein the signal ANALOG corresponds to the output signal of the spot and void filter 500 in the present invention. The character separation logic 405 of the present invention corresponds to the counter 114 shown in FIGS. 1 and 3 of the Butulis application and also to all of the various elements which connect Butulis' counter 114 to his two shift registers 126 and 128 which record the data defining the relative widths of bar and space elements. The decode logic 406 of the present invention corresponds to the read-only memories 130 and 138 shown in the Butulis application, although the logic used by Butulis may be simplified considerably when used in constructing the present invention because the present invention does not necessarily contemplate using start characters ahead of other characters and also because variations in the speed of manual scanning which the Butulis logic must compensate for are not present in the present invention due to the rigid control which is exercised over the speed of scanning by the system clock and array scan logic 401.

The output of the decode logic 406 is a binary representation of the data which is represented by the successive bars and spaces of a character which are scanned. That data is fed into data buffers and error-check logic 600, the details of which are shown in FIG. 6. The data is initially fed into a latch 602 which is driven by a "TAKE DATA" signal derived from logic shown in the Butulis application cited above. The "TAKE DATA" signal is generated only when a set of widths that could conceivably correspond to a valid character has been scanned. During successive scans, the "TAKE DATA" signal is repeatedly generated each time the decode logic thinks that it has again found a valid character. Each time the "TAKE DATA" signal reoccurs, the data previously stored in the latch 602 is advanced into a second latch 604, and the data content of the two latches 602 and 604 is compared by means of a binary comparator 606. A flip-flop 608 tests the result of this comparison. When the outputs of the two latches agree, the flip-flop 608 is toggled into its set state, and it releases a counter 610 so that the counter 610 may begin to count the number of recurrences of the "TAKE DATA" signal which flows to the COUNT input of the counter 610 through a gate 614.

If at any time before the counter 610 reaches full count the data presented by the latches 602 and 604 becomes nonidentical, then the output of the comparator 606 immediately flows through gates 611 and 613 and clears the counter 610 to thereby reinitiate the counting process. Assuming in a specific embodiment that the counter 610 reaches full count at a count of 4, then four successive presentations of the same identical data causes the counter 610 to reach full count and generate output signals which enable a gate 612 to generate a signal A. It should be understood however, that in many applications a single scan of data may be sufficient. The signal A disables a gate 614 and thereby locks the counter 610 at full count to prevent the signal A from terminating prematurely. The signal A also feeds a series of flip-flops 616, 618 and 620 which together generate a "TAKE DATA" pulse and supply that pulse to whatever data capturing device 407 is used in conjunction with the present invention to capture the scanned data -- for example, a tape recorder or computer. Upon receipt of this "TAKE DATA" signal pulse, the device 407 accepts the data presented at the output of the latch 604 as a valid representation of a bar encoded character. It thus requires a sequence of four successive identical scans of the same character before the identification of that character is accepted as valid. With reference to FIG. 8, the presence of large voids and spots within a data character may cause numerous scans to occur during which no valid data is gathered. So long as any area within a bar-encoded character remains free to be scanned vertically at least four times without any of the scans encountering a large spot or void, the character may be accurately read. The adverse effects of small spots and voids are eliminated by the spot and void filter which has already been described.

In FIG. 6, the 0A and 0B signals shown entering the flip-flops 618 and 620 are simply interlaced clock timing pulses which are generated by circuitry disclosed in the Butulis application cited above. The flip-flops 616, 618, and 620 form in combination a simple pulse generator that generates one brief pulse each time the signal A fluctuates positively. Once the flip-flop 616 is set, it enables a 0A clock pulse to set the flip-flop 618 which in turn enables a 0B clock pulse to set the flip-flop 620. An output of the flip-flop 620 clears the flip-flop 616 which then enables a 0A clock pulse to clear the flip-flop 618. The flip-flop 618 then enables a 0B clock pulse to clear the flip-flop 620 so that the pulse-generation is completely returned to its initial state.

The signal A is fed into the character range logic 700 the details of which appear in FIG. 7. The character range logic 700 includes a shift register 702 which is supplied with data from the character separation logic 405 in serial form. This data represents the numerical width of successive bars and spaces that are scanned. In the abovementioned Butulis application, this signal is called the "X ONE" signal, and it appears in FIGS. 1 and 3 of the Butulis application. The "X ONE" signal is fed into the shift register 702 under the control of a CLK 16B data advance signal which may be obtained from FIG. 2 of the Butulis application and which flows to the shift register 702 through the gates 704 and 705. In this manner, the shift register 702 is loaded with binary data representing the actual numeric width of the bars and spaces (or light and dark regions) which have just been scanned.

When the signal A commences, it disables the gate 704 from passing the signal CLK 16B to the shift register 702. The signal A enables the clock pulses 0A generated to flow through a gate 706 to advance the stored data out of the shift register 702 and into a serial-to-parallel converter 708 which is also supplied with pulses from the output of the gate 706. When a complete record of the width of a bar or space has been fed into the converter 708, the counter 710 reaches full count and enables a gate 712 to pass a 0B clock pulse to a latch 714. The latch 714 is interconnected with an adder 716 in such a manner that it adds the width-representing-binary numbers which are successively presented by the converter 708 and thereby calculates the total numeric width of the bar-encoded character which has just been scanned. A second counter 718 counts the number of times the counter 710 reaches full count and determines when the number of widths summed by the latch 714 is equal to the number of bar and space widths in a valid bar-encoded character. The counter 718 then enables a gate 720 to pass a 0A clock pulse that sets a flip-flop 722 and also triggers a latch 724 to accept data. An output of the flip-flop 722 disables the gate 706, stops the advancement of data out of the shift register 702 and into the converter 708, and also prevents further actuation of the latch 714.

At this point, the latch 714 now contains a number proportional to the width of a complete bar-encoded character, including all of its bars and spaces.

When the signal A went high, the trailing edge of a bar-encoded character had just been scanned, and the array-address counter 402 contained a count which may be taken as an indication of the position of the trailing edge of that bar-encoded character which was just scanned. This count value was stored in a latch 726 by the onset of the signal A. An adder 728 now subtracts the binary number contained within the latch 714, representing the width of the character just scanned, from the number within the latch 726, representing the position of the trailing edge of the character just scanned. The resultant difference is stored in a latch 724 in response to the pulse 0A which is permitted to pass through the gate 720 when the counter 718 reaches full count. In this manner, the latch 724 is loaded with a numeric value that indicates the approximate position of the leading edge of the bar-encoded character just scanned, and the latch 726 remains loaded with the number indicating the approximate position of the trailing edge of the bar-encoded character just scanned. Comparators 730 and 732 are now arranged to compare the numeric values within the latches 726 and 724 to the number presented by the array address counter 402 and to generate signals which are combined by an AND-gate 734 that is enabled by the flip-flop 722 to generate a signal that goes high at times when the scan is proceeding across the region where the bars and spaces of the character just scanned were encountered. In this manner, the output signal of the gate 734 may be used to cause other logic elements within the present invention to ignore all scanning signals other than those occurring while the scanning is proceeding directly over a bar-encoded character or the space separating two such characters.

The significance and importance of this arrangement becomes apparent when one examines FIG. 8. The scanning covers not just the bars and spaces of the characters at 801 and 806 but also the printed letters 802, 807, etc. The decode logic 406 can be programmed to tell the difference between the scanning of the bar-encoded characters at 801 and the scanning of the printed letters at 802, so the fact that the letter is scanned causes no difficulty when the bar-encoded character is decoded. However, it is necessary after having accurately captured such a character to detect the wide bar 804 and the wide space 805 which separate that character from the next character. There are portions of the letter A at 802 which could easily be confused with the wide bar 804 and the immediately-following wide space 805. Hence, the output signal of the gate 734 causes the search for a wide bar followed by a wide space to be conducted only using the output of the photodetector elements that are scanning the area directly over the bar-encoded regions and the spaces between such regions. Then the printed letters cannot give a false indication that the end of a bar-encoded character has been reached.

The output of the gate 734 is used to enable a pair of gates 736 and 738 at times when the bar-encoded characters or the spaces between such characters are being scanned, and only at those times. The output of the latch 714 representing the height of a bar-encoded character is divided by four at 716 to give a numeric value proportional to one-fourth the height of a bar-encoded character. This division may be accomplished by transposing the position of the signal bits which lead from the latch 714 by four positions -- thus, there may be no logic within the divide-by-four box 716, but only a shifting of signal lines by four positions. The resulting numeric value is fed into a comparator 740 which compares this value to the output of a counter corresponding to the counter 114 in the above-mentioned Butulis apparatus, which contains a count value proportional to the width of each scanned bar and space. If the numeric value within the counter 114 is greater than that presented by the divide-by-four register at 716, then it may be assumed that the scanning has proceeded until it is over the vertical bar 804 or the adjoining space 805. Hence, the signal output of the comparator 740 (signal B) goes high. This signal B is fed to the gates 736 and 738. The gate 736 is only enabled when a BLK signal is generated by the decode logic 406 (a suitable BLK signal is generated by the flip-flop 214 shown in FIG. 2 of the Butulis apparatus). Hence, when a wide black bar is encountered within the scanning range of the bar-encoded characters, the gate 736 is fully enabled to set a flip-flop 742. The flip-flop 742 then partially enables an AND-gate 744. It also supplies an enabling signal to an AND-gate 738 which initiates a similar search for the wide space 805 (FIG. 8). The absence of the signal BLK accompanied by the presence of the signal B generated by the comparator 740 and the enabling signal from the gate 734 fully enable the gate 738 to set a second flip-flop 746 to signal the scanning of a wide space. The flip-flop 746 then fully enables the gate 744 to supply an input signal to the J terminal of a flip-flop 748. A CB and a CA pulse (these pulses are staggered with respect to one another -- See FIG. 2 of the above-mentioned Butulis application) cause the flip-flop 748 to be set and then cause a flip-flop 750 to be be set immediately thereafter. The setting of the flip-flop 750 clears all of the flip-flops 722, 742, 746 and 748. The next CA pulse then clears the flip-flop 750, since the flip-flop 748 has been cleared.

The signal pulse C generated by the flip-flop 750 is fed back into the data buffers and error check logic 600 (FIG. 6) where it clears the latches 602 and 604 and also clears the flip-flop 608. The signal C thereby reinitiates the search for valid character data in the next bar-encoded character which is to be scanned.

It may thus be seen that the data buffers and error check logic 600 supply successive data characters to the data capturing device 407 accompanied by TAKE DATA pulses each time the code for a new character is presented. Only one such pulse is supplied for each character captured. The entire arrangement disclosed is thus able to decode a series of bar-encoded characters and supply the decoded characters to the data capturing device 407 with a high degree of accuracy and with a low probability of either error or of completing a single horizontal swing of the stylus without capturing a complete set of data. Because the photodetector array is longer in the vertical direction than are the images of the characters which are being scanned, it is possible for the scanner 101 shown in FIG. 1 to be shifted vertically up or down over a considerable distance and to be tilted or skewed throughout a considerable angle without causing an erroneous or aborted scan. It is also possible for the horizontal speed of the scanner 101 to vary a great deal, even in the middle of a scan, without causing an erroneous scan to take plate at least so long as the scanning motion does not reverse itself. Even though the bar-encoded characters are much smaller than those which have been used previously, the present invention is still able to capture data encoded therein more accurately and much faster than was possible with prior arrangements. Data capture is almost always completed with a single pass of the stylus over the characters rather than by multiple passes of the stylus.

While the preferred embodiment of the present invention has been disclosed, it will be understood that numerous modifications and changes will occur to those skilled in the art. It is, therefore, intended by the appended claims to encompass the true spirit and scope of the invention claimed.