JPS5872272A - Optical character reader - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an optical character reading device (hereinafter referred to as r00RJ), and particularly improves the accuracy of cutting out characters on a form.
'Relating to a.

As shown in FIG. 1 (4), in conventional OOR, character cutting is performed using line marks 102 provided for each line, for example, in order to cut out each character when reading characters on a form 101. Based on the starting position of the corresponding line XI! 41[
There is a method of cutting out 1★ character by 103 characters, and a method of cutting out each line based on the starting position of each line, X number 10, Y coordinate 105, using the top or left side of the form 101 as the reference edge without setting a line marker. It is carried out by

However, in actual OOR, due to paper feeding accuracy,
Diagonal forging of the form as shown in No.
Since the apparent frame LOT is shifted from the actual character frame 106, for example, the comma Idt108tj written below the character frame 106 will be placed outside the apparent frame 107, and the first character There was a problem that the image was converted into a space and pgm.

The conventional solution to this problem is to increase the range of the number of characters inscribed by the actual height of the frame. In other words, as shown in FIG. 2 (2), the character capture area 202 is moved 1 part from the actual character frame 201.
It is a good idea to keep it in a large pot. As a result, even if the actual character frame 201 is shifted downward from the expected position due to the skew school, the capture area 201
2 can sufficiently cover the comma symbol within the character frame 201. The same thing can be said about the horizontal direction. The problem with the above method is when the vertically adjacent character frames are close together, that is, when the clear area between the lines is accumulated due to the amount of deviation due to the skew This is a case where the elm price is on the rise and has a small decline. In this case, as shown in FIG. 2 (6), the character capture area 203 on the lower line also includes a part of the upper line (I). For example, a problem arises in which it is mistakenly recognized as abostrophe. What is more important is that the same character (comma) read once in the upper line is read again in the lower line, making the output information difficult for FiOcR users to understand and requiring a lot of effort to correct. The point is that it becomes . In addition, the danger of reading the same character twice is the same in the horizontal direction, and there is a risk that a character already recognized in one digit will be corrupted in the next digit. was devised in view of the above circumstances, and its purpose is to eliminate the sameness such as one statement in the conventional oon, and to read the same character twice even if the line spacing is close, such as rero. The objective is to provide OOR that prevents misreading.

The above object of the present invention is to provide an optical character reading device≠6,
Memorize the coordinates of a specific position of a character cut out in a character frame in a certain line, and in the case of vertical direction, the character in the corresponding frame in the adjacent line below, and in the horizontal direction, in the adjacent frame of the same line. In the character short series ψ, compare the coordinates of the specific position of the character with the memory contents, and if they match, the character is edicted to a.
This is achieved by configuring it to be excluded from the llIN event.

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

FIG. 3 shows character frames on a form. On the @ form 301, character frames of one line and j digits horizontally are arranged at equal intervals. FIG. 4 shows a state in which t-m is determined when two lines of patterns are taken in from one vote made in this way, and this is stored in the memory. Figure 114 shows, for example, the memory in the lower left corner of the circumscribed rectangle 401 of the character component (comma symbol in this case) cut out within the frame of the first digit of the first line on the book IK 301 shown in Figure 3. This shows the situation in which the coordinates (X□1.Y,) are determined, and the thus determined coordinates (X, Y) are 11! Register it as a check mark for the character frame jij in the jth column of the li line.
As shown in the figure.

Based on the table shown in Figure 6, the next row (for vertical orientation)
Compare the check coordinates for the corresponding characters in the frame of t or the next digit (in the case of I [direction) with the check coordinates of the first row and first digit, and as described above, if some of these are If so, the corresponding characters in the frame in the next column across the next line are considered to have been processed.

FIG. 6 is a block diagram of OOR showing one embodiment of the present invention. In the figure, 601 is a form, 602 is a lens, 603 is a sensor, 604 is an amplifier, 606 is a binarization circuit, 606 is a pattern memo IJ, and 607 is a single character extraction logic. Also, the part marked with a broken line M in the figure is a characteristic part of the device of this embodiment, and is a shift register 608.
7 lip/70 tube 609, register 610, 611,
Comparator 612, counter 613, gate number 61, 7 lip/7 g tube 615, counter 616, register 611
7, gates 620, 621, 622, etc.

The form 601 is conveyed to the multi-reading station section by a ψ paper feeding mechanism (not shown), and one scanning portion of the form 601 is sequentially transferred to the seven-dimensional scanner 60 via a lens 602.
Projected to 3. The output of the eclipse sensor 603 is output from the amplifier 60.
4) is converted into black and white binary values by the binarization circuit 606) and stored in the pattern memory 606 &: 2 lines each. The pattern in the I area (as mentioned above, the fruitful character frame! 11 times larger) is extracted. #'i is sent to the shift register 608 one scan at a time, where it is shifted one bit at a time), and if there is a black component, 7 lips and 70 bits 609 are set. Please note that N#
The 7-rib/7-ship 609Fi shown in the figure is reset at the beginning of one scan by a single control logic.

On the other hand, a counter 61 counts each bit in one scan.
3, when the 7 lip/70 lip 609 is written, that is, when the black component is found, it is stored in the register 611.

At the end of one scan, the contents of the register 611 are compared by a comparator 612 with the minimum value up to the previous scan, that is, the black position at the far left, and the current value is smaller than the value in the previous scan. In this case, the current value 1 is selected and becomes the manual input of the gate 620, and at the same time is inputted and stored in the register 610.

The output of id era flip 70 tube 609 is game) 61
4 &:, the timing at the end of one scan (logical product t with No. 1) is taken, and synchronized with the phase of each scan by the D type, 7 lip, 70 tube 615.

On the other hand, the output of the counter 616, which is detected by the clock generated every scan, is the change from 1" to "0" of the 7-lip/7-W tube 615, that is, from a scan containing black to only white. is stored in the register 617 at the point of change to the scanning. The above-mentioned check coordinate for the upper character corresponding to the digit, that is, the character in the character frame of the first row and the third column J, is read out, and the picture logic product of this and the current X coordinate is calculated by the gate 620. In addition, the gate 622 performs a logical product between this and the current Y14i mark, and only when the two match, the gate 621 returns "1'".
This time, the component found in the (1+1)th row, first digit is reported to be a processing error in the first row, first digit, and is excluded from the recognition content.

If the gate 621 does not become “°1”, it is a new/new component, and the next time, that is, the (i +, 2)th row,
To check the first digit, 7) X coordinate, YJffl
is stored in the corresponding storage table of the one-character extraction logic.

In the implementation above, an example is given where the character arrangement on the form is the same in both the vertical and horizontal directions, but this is not always necessary, and for example, for a form with the character arrangement as shown in Figure 7. C also has range A where the character arrangement matches both vertically and horizontally (range surrounded by a dashed line) and #iB
(It is possible to make it with two points of greenery.

The above is the vertical direction, that is, the character turtle I between the lines that touch each other.
11 This is an explanation of how to prevent succession, but on the other hand,
That is, it goes without saying that double reading of characters can be prevented by performing the same processing on the am character digits in the same line. Note that if the purpose is only to prevent duplicate reading between instantaneous character digits in the same line, the pattern memory 606till (which stores pattern data for one line) may be used.

As mentioned above, according to this proposal, the coordinates of the determination position of a character cut out in a character frame of a certain line are memorized in the optical character reading device, and in the case of vertical direction, a, @ &. In the case of a character in a corresponding frame on the lower line, in the horizontal direction, in character processing in an adjacent frame on the same line, the coordinates of the specific position of the character are compared with the stored contents, and if they match. If there is G
Since the character L: is excluded from the recognition target, it has the remarkable effect of completely preventing duplicate reading of the same character.

[Brief explanation of the drawing]

Figures 1 (A) and CB) are diagrams showing the reading method in the conventional OOR, Figure 2 (4) and @ are diagrams showing details of the area for reading in the conventional OOR, and Figure 3 is the diagram showing the reading method of the present invention. FIG. 4 is a diagram showing the pattern in the memory in the embodiment of the present invention. FIG. 6 is a diagram without the check coordinate table. FIG. The block diagram in FIG. 7 is a diagram showing another example of a form. 601111M votes, 602: Lens, 003: Sensor,
612 : comparator,
613, all counter, 614, 620, 621, 6
22: Gate. αυ °″: (A) 1st FA □., ■) Figure 2 Figure 3 Figure 4 Figure 5 Figure 6

Claims (1)

[Claims] ■A device for optically reading characters on a form, including a memory portion for storing character images for at least 11 or 2 lines, and a rectangular space circumscribing the cut-out characters. and means for storing the coordinates of a fixed position in correspondence with the character digit, and when a character is cut out, the mark of the character is compared with the coordinates of the character of the corresponding digit in the already processed adjacent line. and, if these characters are the same, the character is excluded from recognition as a character that has already been processed. - a storage part for storing character images for at least one line, and a means for storing coordinates of a specific position in a rectangular space circumscribing a cut-out round character in correspondence with the character digit; The coordinates of the character are compared with the one mark of the character in the adjacent digit that was trapped, and if they are the same, the scissors character is excluded from recognition as a turtle that has already been caught. An optical character reading device characterized by: