GB1602084A

GB1602084A - Apparatus for reading and analysing bar code characters

Info

Publication number: GB1602084A
Application number: GB20852/78A
Authority: GB
Original assignee: International Business Machines Corp
Current assignee: International Business Machines Corp
Priority date: 1977-06-15
Filing date: 1978-05-19
Publication date: 1981-11-04
Also published as: FR2394856B1; IT7823832A0; IT1096246B; SE434892B; JPS546425A; DK267678A; FR2394856A1; SE7706911L; ES474778A1; ES470764A1; DE2819580A1

Description

(54) APPARATUS FOR READING AND ANALYSING BAR CODE CHARACTERS (71) We, INTERNATIONAL BUSINESS MACHINES COR PORATION, a Corporation organized and existing under the laws of the State of New York in the United States of America, of Armonk, New York 10504, United States of America, do hereby declare the invention for which we pray that a patent may be granted to us, and the method which it is to be performed, to be particularly described in and by the following statement:- In modern information processing it is a normal practice to read data from a document and to transfer this data directly to a data processing system. In bank applications such documents can be bank cheques on which data is coded as magnetic records. These magnetic records can be formed as a bar code, for instance as the well known CMC 7 code.It is a preferred procedure to have these cheques read by a magnetic character reader and after being read the data is automatically transferred to a data processing system.

One problem encountered in reading data on a document and especially in connection with reading a magnetic bar code is the difficulty of correctly decoding the data read. According to Swedish Patent No.

328431 sensed CMC 7 coded data is decoded by temporarily storing the bar code signals and by comparing these signals with a reference signal. This method makes the detected bar code data less sensitive to variations in the feeding speed of the document.

Another method of eliminating errors in decoding is to increase the sensitivity of the document reading elements and the amplifying elements connected to the reading elements. This will however produce a rather expensive apparatus.

UK Patent No. 1452661 discloses another method of improving decoding in connection with automatic character reading. Here the read character pattern is compared with two different reference patterns. If either of these reference patterns is equal to the read pattern, then the character pattern will be accepted.

A drawback of these known systems is that they will be too expensive or they will not provide sufficient reliability in decoding coded records i.e., the number of nonaccepted documents will be too large.

It has been found that when reading magnetic bar codes there will be some typical errors. For instance when a magnetic head reads a bar it can happen that two output pulse signals are provided instead of one pulse signal. Further the document reading speed can vary because of frequency variations in the supply which will influence the document feeding synchronous motor. The feeding speed can further vary because of the nonprecision of mechanical parts used in the mechanical feeding device. This can for instance be the result of using a less expensive rubber drive belt instead of a more expensive cog belt.

Such typical errors are difficult to compensate when decoding using prior art systems.

The object of the present invention is to provide an improved apparatus for reading and analyzing characters represented on a document in accordance with a bar code.

The invention provides apparatus for reading and analysing characters represented on a document in accordance with a bar code, comprising means for transversly scanning said characters to produce signals representative thereof, each character being represented by a group of signals containing information corresponding to the respective transverse distances between corresponding positions on successive bars of the character, means for analysing each group of signals in successive steps, each step utilising different analysis criteria, the analysis criteria of the first step being based on the assumption that there are no errors in the group of signals being analysed and the analysis criteria of each successive step being based on the assumption that there is a respective different type of predetermined error in the group of signals, and means for interrupting said successive steps of analysis when a step of analysis indicates a valid character has been read.

In an order that the invention may be more readily understood, an embodiment thereof will now be described by way of example with reference to the accompanying drawings, in which: Figure 1 is a block diagram of apparatus embodying the invention, Figure 2 is an example of two typical characters represented in the CMC 7 code, Figure 3 is a functional description of the counter and the ring buffer of the apparatus of Figure 1, Figure 4 is a timing diagram disclosing various sensing signals, Figure 5 is a flow diagram of successive analysis steps, Figure 6 is a detailed logic diagram of the counter in the apparatus of Figure 1, and Figure 7 is a timing diagram for the counter illustrated in Figure 6.

The apparatus illustrated diagrammatically in Figure 1 is used to read characters on a document 6 which is provided with magnetic records representing characters coded according to a bar code such as the CMC 7 code. A synchronous motor 11 can, by means of a driving device 12, feed the document into a reading station comprising a magnetic read head 10. The magnetic head is connected to an amplifier 9, which is connected to a pulse distance counter 8. A clock 33 is connected to the counter 8. The counter 8 is, by means of a data line 22, connected to a direct memory access circuit 7, which further is connected, by means of another data line 23, to a memory 2. The memory 2 comprises an input buffer 18, a register field 19 and an output buffer 20. The memory 2 is by means of a line 24, connected to a microprocessor 1, which is connected by means of a line 25 to a microcode storage 3.The microprocessor 1 is further connected by means of a transmission line 21 to input circuits 15 and to output circuits 14 and still further by means of a data line 37 to interface circuits 26. The interface circuits 26 are connected by means of a data line 38 to a terminal loop adapter 27, which is connected to a data transfer loop 36.

The input and output circuits 14 and 15 are connected to a control table 16, which comprises switches 28 and 29, and indicator lamps 30, 31, 32 and 35. The input circuits are further connected to a sensing photo cell 13. The output circuits 14 are further connected to a start relay 34 and to a clock 17, said clock being further connected to said input circuits 15, and to the microprocessor 1.

The apparatus illustrated in Figure 1 will now be explained with reference to Figures 2-5.

A document 6, provided with magnetic records in a bar code, preferably the CMC 7 code, is manually presented to the input of the reading station. When the photodetector 13 detects that the document has been placed into starting position, a signal is transferred from the photodetector 13 through the input circuit 15 and the line 21 to the microprocessor 1. In synchronism with the clock signal from a clock 17 presented as an interrupt signal to the microprocessor 1, the microprocessor 1 will provide a start signal on the line 21 to the output circuits 14 and further to the start relay 34. The start relay 34 will then connect a synchronous motor 11 to an AC power source. This will activate the document feeding device 12 so that the document 6 will be fed into the reading station.The magnetic records will then first pass an energizing station 39 comprising permanent magnets which will activate the magnetic records. Then the magnetic records will pass the magnetic read head 10, which will sense the recorded magnetic bars.

Figure 2 illustrates two typical CMC 7 coded characters, i.e., characters 5 and 6.

Each character comprises 7 vertical bars which are separated by means of 6 spaces.

These spaces represent two long distances between successive bars and four short distances. A short distance can for instance be 0.3 mm and a long distance 0.5 mm. The bar width is preferably between 0.10 and 0.19 mm. When the magnetised bars are passing the magnetic read head 10 the data represented by the bars will be sensed. The read out signals from the magnetic read head 10 will be transferred to an amplifier 9.

The amplifier 9 is conventional and is not a direct part of the apparatus according to the present invention. The amplifier 9 does not need to be a precision amplifier. Preferably this amplifier comprises a preamplifier, a bandpass filter, a logarithmic compression circuit, a pulse detector and a digital pulse former. Normally the amplifier detects all bar code signals from the magnetic head and generates an output signal for each bar.

The output signal from the amplifier 9 will be transferred to the pulse distance counter 8. The pulse distance counter 8 will evaluate the distance between successive bars on the document.

The pulse distance counter 8 operates in such a way that the highspeed clock 33 continuously provides pulses to the counter and steps the counter forwards. When a signal appears from the amplifier 9 the counter will stop. Then the count value in counter 8 will be transferred over the line 22 to the direct memory access circuit 7 and further over the transfer line 23 to the ring buffer 18 in the memory 2. The direct memory access circuit 7 will provide the direct data transfer from the pulse distance counter 8 to the memory 2 without the need of the microprocessor 1 to control this transfer by means of program interrupt.

This direct memory access method is conventional and operates according to a cycle steel principle when the microprocessor 1 has first initiated this transfer routine. This means that the detailed logic of the direct memory access circuit is not an essential part of the apparatus according to the present invention.

As illustrated in Figure 3 it can be seen that before any bar is sensed the pulse distance counter 8 will be stepped by means of pulses from the clock 33 to a maximum count value FF. When a first bar is sensed on the document a transfer of the count value FF through the direct memory access circuit 7 to the first position of the ring buffer 18 will occur and also a reset of the counter value to zero. Then the clock 33 will feed pulses into the counter 8 until a new bar 2 will be sensed. It is supposed that the distance between bar 1 and bar 2 is a short distance e.g., with the value 30. This value 30 will be transferred to the second position of the ring buffer and then the counter 8 again is reset. Then a long distance occurs to the following bar 3 and it is supposed that this will provide a count value 50.The distance between bars 3 and 4 is again supposed to be a short distance and the counter 8 counts up to a value 31.

According to this method the distances between successive bars are step by step brought into the memory 2.

As an example the document feeding speed is 400 mm per second, the clock 33 operates with a frequency of 64 KHz and the clock 17 with a frequency of 0.4 KHz.

The microprocessor 1 analyzes by means of the stored microcode in the microcode storage 3 the information in input buffer 18.

When the microprocessor in this way has analyzed the decoded character data in the input buffer 18 the data will be transferred to the output buffer 20. The output buffer 20 operates also as a ring buffer. The input buffer 18 will be reset after its bare code data has been analyzed and decoded and transferred to the output buffer 20. From the output buffer 20 the data is transferred through the microprocessor I to interface circuits 26 over the transfer line 37 and further over another transfer line 38 to the terminal loop adapter 27. From this adapter the data will be transferred to the loop 36 for further use in a central processing system.

According to Figure 4 it can be seen that a succession of code bars A, B, C, D and E can produce various types of signals from the magnetic read head 10 and the amplifier 9. In the first case which is shown on row 1 the amplifier 9 provide a single output signal coinciding with the trailing edge of each bar. This is the normal case when the reading occurs in a correct way. Row 2 illustrates the case in which each bar sometimes provides double output signals.

This is the case for the bars A, B and D.

Row 3 illustrates the case in which each bar provides double pulses but sometimes provides only a single output pulse coinciding with the leading edge of the bar.

Row 4 illustrates the case where both the leading edges and the trailing edges for the bars provide output pulses.

The microprocessor 1 analyzes the data in the input buffer 18 using criteria according to the various alternatives that are shown in rows I 21 in Figure 4. By means of the microcode stored in the microcode storage 3 the microprocessor will initially analyse the character data in the buffer 18 to see if it is the same form as the data in row 1. If this is the case the character will be considered to be a valid character and will be decoded directly and transferred to the output buffer 20. If the data is not the same form as the data in row 1 this first analysis will not provide a valid result. The microprocessor will then analyse the character data to see if it is the same form as the data in row 2, i.e., when there are two pulses the latter one is valid.This means that when analysing using criteria based on the form of row 2 the first one of two adjacent pulses will be deleted in such a way that the short distance between the pulses will be added to the preceding distance. If this will result in valid analysis the character is considered to be valid and is transferred to the output buffer 20. If however the character data is represented in the form of row 3, then the second analysis will not provide a valid result.

When the third analysis is performed the second one of two adjacent pulses will be deleted and this is made by adding the short distance between two adjacent pulses to the following distance. Character data represented in the form of row 3 will then be considered a valid character which will be transferred to the output buffer 20.

The character data pattern according to row 4 is more unusual. It is a combination of the forms of row 2 and row 3 comprising both leading edge and trailing edge pulses and double pulses. It is however possible to make some assumptions and to analyse such a data pattern. This can be made as a fourth analysis step. It is of course possible to find still other analysis steps, which are based on other known pattern combinations. This makes it possible to step by step analyse the data in the input buffer 18 and then to successfully decode bar code data when certain presupposed errors occur in the signals resulting from reading the data.

There is an advantage in the input buffer 18 being a ring buffer since the analysis of bar code data does not occur in synchronism with the transfer of bar code data from the counter 8 to the input buffer 18. When the bar code data for some characters shows a high percentage of presupposed errors, then the analysis will be delayed so that the number of occupied positions in the ring buffer will increase. If then the following distance data are error free, the analysis will proceed faster and the amount of data in the ring buffer still to be analysed will decrease.

In Figure 5 there is shown a flow chart for analysing bar code data in the input buffer 18. Starting with an input analysis the analysis will proceed using analysis criteria based on the assumption that the data is represented in the form of row 1 of Figure 4.

If this step 1 in series 1 of the analysis provided a valid result, the "yes" line in the flow chart will lead to reset of the ring buffer. Then there will be a return to a new input. If however the first analysis results in an invalid result the operation will automatically proceed to the step 2 analysis.

If the step 2 analysis provides a valid result it will lead to the reset of the ring buffer. If an invalid result is provided the operation will lead to the step 3 analysis. According to Figure 5 it can be seen that if the analysis steps according to series 1 have not provided a valid result, the end of the series 1 analysis will lead to the beginning of the series 2 analysis. In series 2 analysis all the same analysis steps 1, 2, 3 etc to N will be performed as in series 1. The difference between the analysis in series 2 as compared with that in series 1 is based on the assumption that the distances between successive bars of the code are shorter than normal. It can for example be assumed that the short reference distance has a value 25 instead of the normal value 30 and that the long reference distance has a value 45 instead of 50.The microprocessor 1 will then use the analysis some distance data with lower value stored in the microcode storage 3. If the series 2 analysis does not provide a valid result, then the analysis proceeds to the series 3 analysis. The series 3 analysis differs from the series 1 analysis in that the distances between successive bars are assumed to be longer than normal, that means that the short reference distance is assumed to be 35 instead of 30 and the long interval is assumed to be 55 instead of 50, which means that distance data with higher values stored in the microcode storage is used.

If the end step of series 3 analysis leads to an invalid result, i.e., it has not been possible to find a correct form of analysis, an error routine will follow. If however the series 3 analysis in any position leads to a valid result, a transfer routine for the reset of the ring buffer will occur.

The advantage of the series 2 and the series 3 forms of analysis is that the analysis of bar code data will be dynamic. This means for instance that some mechanical variations in the speed due to the driving synchronous motor for the document feeding can be compensated. Generally speaking it can be said that the system allows greater tolerances in the mechanical and electrical construction of the magnetic record reading arrangement. A suitable basis for analysis of reference distance data is to add all distance data for a sensed character and then compare the same with the sum for normal distance data in a character. Then new reference data will be established for short and long distances by taking into consideration this comparison value.

As illustrated in Figure 6 the clock 33 is connected to the oscillator for the microprocessor 1 and receives over a line 55 system clock pulses with, for instance, a frequency of one megacycle. These pulses are fed into a binary counter 50. The output of the binary counter 50 is combined in AND circuits 51 and 52 and in inverter circuits 53 and 54 in such a way that new clock pulses with a lower frequency are provided. These new clock pulses are called +PH2 +PHI and -PH1 and are illustrated as outputs in Figure 6.

As also illustrated in Figure 6 it can be seen that distance counter 8 comprises a shift register 57 receiving magnetic readout data from the amplifier 9 in Figure 1 over a line 56. The shift register 57 will be stepped by signals from the clock 33 by using the -PHI signal line.

The shift register 57 operates as a delay circuit having its first two outputs connected to two inverting circuits 58 and 59. Only such a magnetic read out data signal on line 56 which has sufficient duration can switch a NAND circuit 60 so that another NAND circuit 62 and an inverter circuit 63 can be activated. A binary counter 64 counts distances between successively sensed bars. As long as the NAND circuit 60 is activated from the shift register 57 the clock signal PHl will step the binary counter 64 through the NAND circuit 62 and the inverter circuit 63.When the stop signal is received from the shift register 57 through the gate 60 the stepping of the counter 64 will stop due to signals from the gate circuits 62 and 63, and then the inverter circuit 65 and the NAND circuit 66 will reset the binary counter 64, simultaneously with transfer of its content to register 69. In connection with the reset and transfer operation an AND gate 67 will by means of a clock signal PH2 send a calling signal to the direct memory access circuit 7 for receiving the distance data from the register 69.

If the distance between two successive bars exceeds a given higher threshold value (FF) the binary counter 64 will automatically be stopped due to a feedback by means of a NAND circuit 61 which feeds the gate 62.

According to Figure 7 it can be seen that a down going magnet in read out data signal will influence the shift register output so that the NAND gate 60 will be switched after two PHI clock pulses. This occurs in connection with the trailing edge of the third PMl signal. This will result in a call from AND circuit 67, after which the binary counter 64 will start to count. It can further be seen that the short data signal pulse at PHl value 8 will not influence the binary counter 64. However the long data signal pulse at PHl value 12 and 13 will stop the binary counter 64 and initiate a calling signal to the direct memory access circuit.

This arrangement prevents short data signal pulses such as noise signals from stopping counter 64.

In accordance with Figure 1 the operator can set the device into a test mode by means of a switch 29 on the control table. The microprocessor 1 will then initiate over the output gate circuits 14 a number of test signals to amplifier 9. The test result can then be analyzed by circuits 8, 7 and 2 or the output of the amplifier 9 can be sensed directly by the microprocessor 1 over the input gates 15. It is convenient to indicate the test mode by an indicator lamp 32.

The purpose of the indicator lamp 35 is to indicate the availability of the loop 36.

The process indicator lamp 30 is used to indicate that data in buffer 20 is available for sending to the terminal loop 36.

The inhibit indicator lamp 31 is used to indicate that the buffer capacity in the memory 2 is fully used. The motor 11 will then not feed new documents before data from the output buffer is being sent to the outer loop 36.

WHAT WE CLAIM IS: 1. Apparatus for reading and analysing characters represented on a document in accordance with a bar code, comprising means for transversly scanning said characters to produce signals representative thereof, each character being represented by a group of signals containing information corresponding to the respective transverse distances between corresponding positions on successive bars of the character, means for analysing each group of signals in successive steps, each step utilising different analysis criteria, the analysis critieria of the first step being based on the assumption that there are no errors in the group of signals being analysed and the analysis criteria of each successive step being based on the assumption that there is a respective different type of predetermined error in the group of signals, and means for interrupting said successive steps of analysis when a step of analysis indicates a valid character has been read.

2. Apparatus as claimed in claim 1, in which, in error free operation, the scanning means produce a group of pulses representing a character comprising only first pulses at times corresponding to the respective instances at which scanning occurs of the corresponding positions on the successive bars of a character.

3. Apparatus as claimed in claim 2, in which, in a first predetermined type of error, the scanning means produce additional pulses in a group of pulses representing a character at times corresponding to instances other than those instances at which the corresponding positions are scanned and in which the analysing means comprise means for suppressing the additional pulses to provide an error free group of pulses representing the character.

4. Apparatus as claimed in claim 2, in which in a second predetermined type of error, the scanning means produce a group of pulses representing a character comprising first pulses at only some of the instants at which said corresponding positions of the successive bars are scanned and further pulses at instants at which other corresponding positions of the successive bars are scanned, a further pulse being produced for each bar, and in which the analysing means comprise means for suppressing said first pulses.

5. Apparatus as claimed in claim 2, in which, in a third predetermined type of error, the scanning means produce a group of pulses representing a character comprising first pulses at only some of the instants at which said corresponding positions on the successive bars are scanned and further pulses at only some of the instants at which other corresponding positions of the successive bars are scanned and in which the analysing means comprise

**WARNING** end of DESC field may overlap start of CLMS **.

Claims

**WARNING** start of CLMS field may overlap end of DESC **. binary counter 64 through the NAND circuit 62 and the inverter circuit 63. When the stop signal is received from the shift register 57 through the gate 60 the stepping of the counter 64 will stop due to signals from the gate circuits 62 and 63, and then the inverter circuit 65 and the NAND circuit 66 will reset the binary counter 64, simultaneously with transfer of its content to register 69. In connection with the reset and transfer operation an AND gate 67 will by means of a clock signal PH2 send a calling signal to the direct memory access circuit 7 for receiving the distance data from the register 69. If the distance between two successive bars exceeds a given higher threshold value (FF) the binary counter 64 will automatically be stopped due to a feedback by means of a NAND circuit 61 which feeds the gate 62. According to Figure 7 it can be seen that a down going magnet in read out data signal will influence the shift register output so that the NAND gate 60 will be switched after two PHI clock pulses. This occurs in connection with the trailing edge of the third PMl signal. This will result in a call from AND circuit 67, after which the binary counter 64 will start to count. It can further be seen that the short data signal pulse at PHl value 8 will not influence the binary counter 64. However the long data signal pulse at PHl value 12 and 13 will stop the binary counter 64 and initiate a calling signal to the direct memory access circuit. This arrangement prevents short data signal pulses such as noise signals from stopping counter 64. In accordance with Figure 1 the operator can set the device into a test mode by means of a switch 29 on the control table. The microprocessor 1 will then initiate over the output gate circuits 14 a number of test signals to amplifier 9. The test result can then be analyzed by circuits 8, 7 and 2 or the output of the amplifier 9 can be sensed directly by the microprocessor 1 over the input gates 15. It is convenient to indicate the test mode by an indicator lamp 32. The purpose of the indicator lamp 35 is to indicate the availability of the loop 36. The process indicator lamp 30 is used to indicate that data in buffer 20 is available for sending to the terminal loop 36. The inhibit indicator lamp 31 is used to indicate that the buffer capacity in the memory 2 is fully used. The motor 11 will then not feed new documents before data from the output buffer is being sent to the outer loop 36. WHAT WE CLAIM IS:

1. Apparatus for reading and analysing characters represented on a document in accordance with a bar code, comprising means for transversly scanning said characters to produce signals representative thereof, each character being represented by a group of signals containing information corresponding to the respective transverse distances between corresponding positions on successive bars of the character, means for analysing each group of signals in successive steps, each step utilising different analysis criteria, the analysis critieria of the first step being based on the assumption that there are no errors in the group of signals being analysed and the analysis criteria of each successive step being based on the assumption that there is a respective different type of predetermined error in the group of signals, and means for interrupting said successive steps of analysis when a step of analysis indicates a valid character has been read.

means for determining the spacings of the positions on the same bar giving rise to two pulses of the group and means for deriving therefrom spacings corresponding to a valid character.

6. Apparatus as claimed in any one of the preceding claims, further comprising a storage device for storing the signals of a group until the necessary analysis thereof has been completed.

7. Apparatus as claimed in claim 6, further comprising means for removing analysed groups of signals from said storage device and for thereafter inserting further groups of signals to be analysed into said storage device.

8. Apparatus as claimed in any one of the preceding claims, further comprising a microprocessor for controlling and performing said analysis.

9. Apparatus for reading and analysing characters represented on a document in accordance with a bar code, which apparatus is substantiallv as hereinbefore described with reference to and as illustrated in, the accompanying drawings.