JPS6122837B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an optical character reader, and more particularly to an optical character reader that performs scanning using a one-dimensional semiconductor sensor and paper feed to correctly read forms without line marks.

In an optical character reader using a semiconductor CCD sensor, as shown in FIG. , and an image of a line 7 perpendicular to the arrow 6 on the form 1 is projected onto the semiconductor CCD sensor 19 by the lens 14. The semiconductor CCD sensor 19 is composed of, for example, 2048-bit photo-sensing cells, and by inputting a clock, each photo-sensing cell sequentially outputs an electric signal corresponding to the input light to the binarization circuit 20. 2
The signal binarized by the digitization circuit 20 is stored in the memory 21
After being stored in , it is read out and input to the recognition circuit 22 , where the recognized result is sent to the control unit 10 .

The memory 21 shown in FIG. 1 generally has a capacity sufficient to store one line's worth of signals, and stops the paper feeding mechanism after one line's worth of signals is input.

For this purpose, a line mark 8 is printed on the form 1, and when the line mark detector (not shown) detects the line mark 8, a detection signal is sent to the paper feed mechanism, and after one line has traveled, the form is printed. 1 is stopped.

As described above, in FIG. 1, since the line position can be detected by the line mark 8, there is an advantage that even if the form 1 expands or contracts due to changes in temperature and humidity, the effect on reading is small.

However, since it is necessary to print the line marks 8 on the form 1, the cost of printing the form increases and there are significant constraints on the form design.

On the other hand, the distance from the leading edge of the form 1 is programmed in the control unit 10, and by utilizing the fact that the running speed of the form 1 is constant, the time after the leading edge of the form 1 passes the scanning line 7 is determined. It is also possible to detect rows. According to this method, the line mark 8 is not required, but since the traveling speed must be kept constant with high precision, the paper feeding mechanism becomes expensive, and expansion and contraction of the paper due to temperature and humidity cannot be corrected. Note that since the rollers in the paper feeding mechanism are usually made of rubber or the like, the diameter of the roller changes depending on temperature and humidity, and the surface of the roller wears out due to long-term use, causing the diameter of the roller to change.

The expansion and contraction of paper due to temperature and humidity is at most ±2%, that is, the form 1 with a length of 300 mm expands and contracts by ±6 mm. Therefore, in the worst case, the characters written on the rear edge of the 300mm form 1 will be shifted up or down by 6mm, and this value means that small characters, such as type (approx. 2.54mm in height) If it is being printed, etc., it becomes a fatal value and accurate reading becomes impossible.

The purpose of the present invention is to solve such problems by scanning with a one-dimensional semiconductor sensor and paper feed, and when reading data on a form without line marks, it is possible to avoid the expansion and contraction of paper due to changes in temperature and humidity. To provide an optical character reader capable of accurately reading a field without being affected.

The optical character reader of the present invention includes a memory having a capacity capable of storing a binary signal of the entire document, a document detector that detects the document on a scanning line, and a document length measured by the document detector. It is characterized by providing means for correcting the reading position on the form by comparing it with a preprogrammed length of the form.

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 2 is a schematic diagram of an optical character reader showing an embodiment of the present invention.

The form 1 to be read is moved at a constant speed in the direction of an arrow 6 by a paper feed mechanism (not shown). A start signal from the control unit 10 causes the form 1 to start running. When the scanning line 7 is irradiated by the lamp 18, the scanning line 7 is projected onto the semiconductor CCD sensor 19 by the lens 14. semiconductor CCD
The sensor 19 is composed of, for example, 2048-bit photodetection cells, and sequentially outputs analog electrical signals corresponding to the irradiation light of each photodetection cell to the binarization circuit 20 using a clock from a clock oscillator (not shown). For example, the magnification of the lens 14 is set to 8 on the paper surface 1.
When the resolution is set to 1 line/mm, a scanning line 7 with a width of 256 mm is projected within the 2048-bit semiconductor CCD sensor 19. Furthermore, the semiconductor CCD sensor 19 can be replaced with a one-dimensionally arranged photo array or the like.

The binarization circuit 20 performs shading correction on the input analog electrical signal and converts the black signal to "1" and the white signal to "0" by 2, for example, by a slicing circuit.
It is converted into a value, outputted to the memory 21, and stored. The memory 21 can write binary signals, for example, eight signals forming one byte. memory 2
The capacity of 1 is 2048 bit semiconductor CCD sensor 1
9 is used to set the resolution to 8 lines/mm, and the size of form 1 to be read is, for example, A4 size (210 mm x 297 mm).
mm), it is approximately 600K bytes.

Further, the memory 21 can be read at any address by the recognition circuit 22, so the recognition circuit 22 inputs the address in the memory 21 of the field on the form 1 to be read from the control unit 10, and
The binarized character pattern to be read is read from the memory 21, recognized, and the result is output to the control unit 10.

The control unit 10 has a function of controlling the reading method of the form 1 programmed in advance by the user, controlling the paper feeding mechanism, outputting the programmed information to the recognition circuit 22 for each field, and inputting recognition results from the recognition circuit 22. It has functions such as output editing, but these functions can be easily realized by using a microcomputer.

FIG. 3 is an explanatory diagram of the reading position correction method on the form in FIG. 2.

Now, if you scan at the original 8 lines/mm pitch,
If the length of form 1 is 300mm, the entire scanning line is
2400 pieces (Nominal). However, as a result of measuring the passage time of the form 1 by the form detector 30,
When 2640 scanning lines were counted, a 10% elongation was detected, so when reading the first field 2, the 80th line was corrected from the 88th line, and when reading the second field 5,
Correct the 2320th line to the 2552nd line. This correction can be easily executed by the program of the control section 10.

FIG. 4 is a flowchart of a read operation illustrating an embodiment of the present invention.

The contents that the user programs in advance in the control unit 10 include the coordinates of fields to be read on the form, character types, and the reading order of the fields.

First, in step 31, a start signal is sent from the control section 10 to the paper feeding mechanism to cause the form 1 to be read to run at a constant speed in the direction of the arrow 6. Next, in step 32, when the form 1 is run on the scanning line 7 by the paper feed mechanism, the form detector 30 detects this and notifies the control unit 10, so that the binarization circuit 20 The binary signal is stored in the memory 21. The form detector 30 includes, for example, a photodetector such as a solar cell placed below the scanning line 7, and a lamp 18.
This can be easily detected by determining whether the amount of light is blocked by the form 1 or not.

At this time, in step 33, the control unit 10 clears and sets the measurement timer, and at the same time, the memory 2
Start writing at 1. In step 34, it is determined whether the form detector 30 has been turned off.

The control unit 10 determines that the form 1 has passed through the scanning line 7 when the light from the lamp 18 enters the form detector 30 again.

At this time, in step 35, the control unit 10 reads the timer value and at the same time stops writing to the memory 21. Further, in step 39, the running of the form 1 is also stopped. As mentioned above, the memory 21 is
It has a sufficiently large capacity and can binarize and store all the image data on the form 1 input at the specified resolution.

At step 36, the control unit 10 calculates a correction value using the programmed length of the form 1 and the timer value. That is, the form detector 30 detects form 1.
The passing time of Form 1 is measured using a timer, and from this time, using the fact that the traveling speed of Form 1 is constant,
The actual length of the form 1 is calculated and the length of the form 1 programmed in advance is corrected.

At step 40, the controller 10 starts reading from the memory 21. The control unit 10 recognizes the characters of the field to be read according to a preprogrammed order. For example, if field 2 is programmed as the first field to be read first, the location of the programmed field is converted to an address on memory 21.

That is, in step 37, the coordinates of the first field are corrected and the address is converted. Address conversion is performed by correcting the length of the form 1 programmed in advance and the measured length, since the address is stored in the memory 21 in order from the leading edge of the form 1 at a predetermined resolution (e.g. 8 lines/mm). It can be easily calculated by doing this.

Next, in step 38, the character pattern in the memory 21 is read out and input to the recognition circuit 22. That is, the recognition circuit 22 reads the binary signal of the address inputted from the control section 10 from the memory 21, recognizes it, and outputs the result to the control section 10.

At this point, the process returns to step 37 and moves on to correcting the second field (for example, field 5) that is programmed to be read next.

By repeating these operations, any field can be read without being affected by the expansion and contraction of the paper.

In the embodiment, the output of the binarization circuit 20 has been described as a black signal "1" and a white signal "0", but it is also possible to store multi-level digitized signals in the memory 21 depending on the shading. . In this case, it is necessary to increase the capacity of the memory 21.

In addition, in the embodiment, eight binary signals are configured as one byte and input to the memory 21, but two
If an encoding circuit is inserted between the digitizing circuit 20 and the memory 21, and a decoding circuit is inserted between the memory 21 and the recognition circuit 22, and the encoding circuit performs run-length encoding, the capacity of the memory 21 can be compressed. can.

Furthermore, although the form detector 30 is provided in the embodiment, the form 1 can also be detected by monitoring the level of the analog electrical signal from the semiconductor CCD sensor 19 without providing this. This takes advantage of the fact that there is a difference in level between when the form 1 is on the scanning line 7 and when it is not.

In addition, in the embodiment, after all the binarized signals of the form 1 are stored in the memory 21, character recognition processing is performed while correcting the length of the form 1.
In order to improve the throughput of reading the form 1, by adding logic to detect the storage status in the memory 21, the character can be read immediately after the field to be read is stored in the memory 21 without performing length correction. Finally, measure the length of form 1 and compare it with the pre-programmed length of form 1. If there is no expansion or contraction, no expansion or contraction correction is performed, and if expansion or contraction is measured, the length is measured again. It is also possible to perform character recognition while performing expansion/contraction correction. Thereby, the throughput can be improved for reading the form 1 which has a small degree of expansion and contraction.

As explained above, according to the present invention, a memory having a capacity that can store the binarized signal of the entire document is provided, and the expansion/contraction correction of the document is performed after storing the binary signal, so that the document without line marks can be corrected. Any field can be read accurately without being affected by the expansion and contraction of the paper.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an optical character reader using a conventional semiconductor CCD sensor, FIG. 2 is a block diagram of an optical character reader showing an embodiment of the present invention, and FIG. FIG. 4, which is an explanatory diagram of the reading position correction method, is a flowchart of the reading operation showing an embodiment of the present invention. 1: Form, 2, 3, 4, 5: Field to be read, 6: Running direction of form, 7: Scanning line, 8:
Line mark, 10: Control unit, 14: Lens, 18:
Lamp, 19: Semiconductor CCD sensor, 20: Binarization circuit, 21: Memory, 22: Recognition circuit, 30:
Form detector.

Claims (1)

[Claims] 1. A photoelectric conversion means that optically scans a form, detects the reflected light, and converts it into an electrical signal; a binarization circuit that binarizes the output signal of the photoelectric conversion means; a storage means for storing the output; a means for measuring the length of the form by detecting the edge of the form; and a storage means for controlling the operation of each of these means and storing a binary signal for one form. After the length of the measured form is compared with the length of a previously prepared form, the amount of expansion and contraction of the form is calculated, and the reading position of the data stored in the storage means is determined based on the amount of expansion and contraction of the form. 1. An optical character reader comprising: a control means for correcting the position information; and a recognition means for reading data from the storage means and recognizing characters based on the position information corrected by the control means.