JPH0570876B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (A) Field of Industrial Application The present invention relates to a character recognition device that recognizes characters written on manuscript paper, and particularly relates to switching of recognition modes.

(b) Prior art As disclosed in Japanese Patent Publication No. 58-31628, a character entry field on a manuscript sheet is constructed by recording rectangular character frames in a grid pattern, and each line of this character entry field is Conventionally, timing marks are recorded and characters are cut out based on the timing marks.

(c) Problems to be Solved by the Invention When the size of the character frame on the manuscript paper is constant, the correct character can be cut out by using timing marks as in the past. However, when considering the number of characters, there is a demand for using manuscript paper with character frames of different sizes recorded, and in this case, the cutout range of characters etc. will be different, so the recognition mode must be switched.

Therefore, it is conceivable to manually switch the recognition mode according to the size of the character frame, like changing the paper size of a copying machine, but this method makes the operation cumbersome and is prone to incorrect operations. This may lead to a decrease in the recognition rate based on the

(d) Means for Solving the Problems The present invention provides that, for manuscript papers in which the size of the character frame of the character entry field is different, timing marks at predetermined positions are recorded with different lengths, and timing marks at predetermined positions are recorded with different lengths. The apparatus is provided with a detection means for detecting the length of the timing mark at the timing mark, and a mode setting means for setting the recognition mode according to the detection result.

(e) Effect In the present invention, the length of the timing mark at a predetermined position on the manuscript paper is detected, and by this detection, the size of the character frame of the input manuscript paper is identified. Since the recognition mode is set according to the detection result, when a document sheet with a character frame of a different size is input, the recognition mode is automatically switched to the corresponding mode.

(F) Embodiment The format of a manuscript paper according to the present invention is shown in FIGS. 1 and 2, and a block diagram showing the structure of a character recognition apparatus according to the present invention is shown in FIG.

Manuscript sheets 1 and 2 in Figures 1 and 2 are both A4 size sheets, and character frames forming character entry fields are recorded in a grid pattern, and each line has character cutouts. Timing marks for each are recorded.

However, the sizes of the character frames 3 and 4 of the manuscript papers 1 and 2 are different, 10 mm square and 8 mm square, respectively, and in order to identify them, the lengths of the timing marks 5b and 6b on the second line are different. is attached.
That is, on the manuscript paper 1, the length of the timing marks 5b, 5c, etc. from the second line onward is 4 mm, and on the manuscript paper 2, the length of the timing mark 6bn on the second line is 10 mm, and the length of the timing marks 6bn on the second line is 10 mm, The length of the timing marks 6c, 6d, . . . after the second one is 4 mm. In addition, the timing marks 5a and 6a on the first line
The length of both sheets is 10 mm in order to locate the beginning of manuscript sheets 1 and 2.

As shown in FIG. 3, the character recognition device includes a reading section 7, a buffer memory 8 for storing read data,
Character cutting section 9, recognition section 10, paper feeding mechanism section 1
1, paper feed control section 12, control section 13, operation section 1
4. Consists of a timing mark detection section 15 and a mode setting section 16, and the reading section 7 has a photoelectric conversion element that is slightly longer than the width of the manuscript papers 1 and 2 in the horizontal direction and 0.1 mm in the vertical direction. A fixed line sensor is provided that reads the range at once.

Therefore, when a start command is given from the operation unit 14, the paper feed control unit 11 drives the paper feed mechanism unit 12 to
First, original paper 1 or 2 is fed until the line a reaches the position of the line sensor, and the beginning of the paper is found. Next, the paper feeding mechanism section 12 feeds the paper by a predetermined amount, for example, 17 mm, corresponding to about 1.5 lines of the manuscript paper 1, and during this time, the reading section 7
reads the data in the range shown by the dotted line in FIGS. 1 and 2, and stores the read data in the buffer memory 8.

As shown in FIG. 4, the timing mark detection unit 15 detects the X address of the right end of each timing mark.
An address detection unit 17 that detects X _R and the lower end Y address Y _U , a length detection unit 18 that detects the length l of each timing mark, and a length determination unit that compares the detected length l with a predetermined value. 19, the length determining unit 19 compares the length l of each timing mark with a predetermined value l _L smaller than 4 mm, for example, 3.5 mm.
Comparison section 20 and input manuscript paper 1 or 2
For only the timing mark 5b or 6b in the second row, the second comparing section 21 is provided to compare the length l of the mark with a predetermined value _lH larger than 4 mm, for example, 5 mm. When the result of the first comparison unit 20 is l>l _L , it is determined that the detected mark is not a noise such as dust but a correct timing mark, and the detected X and Y address
X _R and Y _U are sent to the character cutting section 9, and the comparison result of the second comparing section 21 is sent to the mode setting section 16.

By the way, the buffer memory 8 contains the 17
As a result of the paper feed of mm, read data corresponding to the timing mark 5a or 6a of the first line and the timing mark 5b or 6b of the second line of the document paper 1 or 2 is stored, and therefore, the timing mark detection unit 15, the comparison result of the second comparison unit 21 is l/l _H when manuscript paper 1 is input, and l>l _H when manuscript paper 2 is input. Become. Then, the mode setting unit 16 sets the recognition mode to the first mode when the comparison result is l< _lH , and sets the recognition mode to the second mode when the comparison result is l> _lH .
Set to mode.

In other words, in the first mode, the character cutting range is 10mm.
Set the range E ₁ corresponding to the character frame, and set the paper feed amount to P ₁ . In addition, in the second mode, the character cutting range is set to E2, which is narrower than _E1 , _which corresponds to an 8 mm character frame, and the paper feed amount is set to be smaller than _P1.
Set to _P2 .

Therefore, the character cutting section 9 extracts 1 based on the addresses X _R1 and Y _U1 of the timing mark of the first line detected by the timing mark detecting section 15 and the cutting range E ₁ or E ₂ set in each mode. The character pattern data included in the row is cut out character by character and output to the recognition unit 10. The recognition unit 10 recognizes each character through preprocessing, feature extraction, and standard pattern matching. During recognition of the first line, the paper is fed by the amount set in each mode, and the pattern data of the second line is stored in the buffer memory 8. Thereafter, while the previous row is being recognized in the same way, the pattern data for the next row is stored in the buffer memory 8 by time-sharing processing. Then, timing marks are detected for each line, and the addresses X _R and Y _U of the timing marks for each line are determined. Based on these addresses and the set cutting range, characters are sequentially cut out and recognized. Ru.

In this way, in this embodiment, depending on the length of the timing mark on the second line of the manuscript paper, it is possible to determine whether the input manuscript paper is a paper with a 10 mm square character frame or a paper with an 8 mm square character frame. The recognition mode is automatically switched to the first mode or the second mode.

Note that if the capacity of the buffer memory 8 is small, about two lines, the newly read data must be written to the memory address of the read data that has finished being recognized. The feed amount must also be changed, but when the capacity of the buffer memory 8 is large, the first mode and the second mode
It is not necessary to change the paper feed amount depending on the mode.

(g) Effects of the invention According to the present invention, just by inputting the manuscript paper,
Since the recognition mode is automatically switched, operability is improved and correct character extraction is reliably executed.

[Brief explanation of the drawing]

1 and 2 are diagrams showing the format of manuscript paper according to the present invention, FIG. 3 is a block diagram showing the configuration of the character recognition device according to the present invention, and FIG. 4 is a block diagram showing details of the timing mark detection section. It is. Explanation of main figure numbers, 1, 2...Manuscript paper, 3, 4
...Character frame, 5a to 5d, 6a to 6d...Timing mark, 8...Buffer memory, 9...Character cutting section, 15...Timing mark detection section.

Claims (1)

[Claims] 1. In a character recognition device that uses manuscript paper on which timing marks for character cutting are recorded for each line of the character entry field, for manuscript papers in which the character frame of the character entry field has a different size,
A detecting means for recording different lengths of the timing marks at predetermined positions and detecting the length of the timing marks at the predetermined positions; and a mode setting means for setting a recognition mode according to the detection result. 1. A character recognition device, characterized in that the recognition mode is automatically switched according to the input manuscript paper.