JPS59206987A - Letter recognizing device - Google Patents

Abstract

PURPOSE:To improve accuracy of letter segmenting even when fluctuation occurs in carrying speed etc. by correcting position of segmenting of other letter line according to the actual position of segmenting of letter segmenting of a continuous letter line. CONSTITUTION:A control section 4 receives reading control information (FC) from an external control device 5. A sheet of form is fed and reading pattern is stored in a buffer memory (BM) 2. Address of each letter position of a letter line to be recognized is calculated from the FC, and converted to addresses CA1, CA2-CAn in the BM2. The BM2 is scanned from address a specified value (e) prior to reading start position MS1 given by the FC, and detection and segmenting operation of letters are started. In detecting and segmenting of letters, if a black part is detected by scanning of the BM2, it is judged whether the black part is a part of a letter or not. If it is a letter, segmenting the first letter is performed as the beginning of the first letter.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a character recognition device, and more particularly to a character recognition device that can reliably perform character recognition even when there are fluctuations in the conveyance speed of documents.

[Technical background of the invention and its problems]

In conventional optical character recognition devices, there are many cases where characters are cut out and recognized incorrectly due to fluctuations in the conveyance speed of the document, especially in document-type optical character recognition devices that carry the document while holding it perpendicular to the conveyance surface. However, due to limitations in transport technology, fluctuations in transport speed were a major problem. When the conveyance speed fluctuates greatly, it is particularly difficult to cut out and recognize blank spaces (nupaine). This is because it is difficult to distinguish between a continuous white area due to speed fluctuations and a continuous white area due to NuPace.

[Purpose of the invention]

The present invention has been made in consideration of the above circumstances, and it is an object of the present invention to provide a character recognition device that can correctly cut out and recognize characters even when there are fluctuations in the transport speed of a form.

[Summary of the invention]

To achieve this objective, the character recognition device according to the present invention comprises:
Cutting out one character pattern at a time for continuous character lines with no blank spaces in the reading pattern of the form, calculating the average positional error between this cutting position and a predetermined temporary cutting position,
The present invention is characterized in that the predetermined temporary cutting position # for the character line of the character line is corrected by the average position error to cut out the pattern of the character line of the character character line.

[Embodiments of the invention]

FIG. 1 shows a character recognition device according to an embodiment of the present invention.

The photoelectric conversion unit 1 reads a form containing characters, etc., and
Change to @. The converted reading pattern for one form is stored in the buffer memory 2.

The control section 4 has the ability to cut out the reading pattern stored in the buffer memory 2 one character at a time and controls the entire character recognition device. The character recognition unit 8 recognizes the character-by-character pattern cut out by the control unit 4.

The character recognition device according to this embodiment reads and recognizes a form with multiple lines written at once.
It is assumed that at least one line out of the plurality of lines is written with continuous characters, etc., with no spaces in the middle. Here, explanation will be given assuming that such a continuous character line exists in the first line of the form, for example.

Next, the operation of this character recognition device will be explained with reference to FIGS. 2 and 8. First, the control unit 4 receives reading control information (hereinafter referred to as "FC") such as character line position, reading start position, character type, etc. of the form from the external control device 5 (block 11). Next, one sheet is fed, and the reading pattern of the entire sheet is stored in buffer memory 4 as shown in FIG. 2 (block 12).
). Next, the control unit 4 scans the buffer memory to detect the left end address yf of the form (block 13). Further F
Calculate the address of each character position in the character line to be recognized from C, add the left end address of the form obtained earlier, and obtain the addresses OA, , OA, , . . . in the buffer memory 2.

Convert to OAn (n is the number of characters) (block 14).

Next, in block 15, it is determined whether the character line to be recognized is the first line, and if it is the first line, the process moves to block 16. In block 16, the reading start position tMs given by the FC
, the buffer memory 2'Ik is scanned from an address a predetermined value earlier than fffe, and character detection and extraction operations are started.

This predetermined straightness e is determined by a form conveyance speed error, form printing ejection, and the like. Character detection and extraction operations involve scanning the buffer memory 2 to detect a black part, determining whether the black part is part of a character, and if it is a character, starting from the beginning of the first character. The first character is extracted as follows.

Cut out the second and subsequent characters in the same manner. after that,
Address TAL of the center position of each character actually cut out
# T'* I..., Tl1kn and the address Ok1. which was calculated earlier. The error "1*dl e "'m dyI with respect to OA, .

In the case of FIG. 2, it can be seen that the calculated address OA and the actually extracted address TAi are shifted to the right. Next, after storing the average @D' in a predetermined memory (block 17), the extracted characters are recognized (
Block 18). Next, in block 20, it is determined whether there is a next line, and if so, the process moves to block 13 to perform character recognition on the next line.

To recognize the next line of characters, first calculate the calculated address OA 1f for each character position according to the FC (blocks 13 and 14).
Since it is not the line, the processing is moved to block 18 (block 15), where the previously determined average position error Djjy- is added to the address CLt of each character position to correct it and cut out the character. Next, the cut out characters are recognized (block 19). If there is another line, the process moves to block 13 (block 20) and the above process is repeated.
If recognition has been completed for all character lines in the form, the process moves to block 12 (block 20),
Repeat the above process for a new form.

In this way, according to the present embodiment, the cutting position of the continuous character line is corrected in accordance with the actual character cutting position of the continuous character line, so even if there is a change in the conveyance speed of the form, etc., the cutting position can be corrected. You can cut out characters.

Note that character detection and extraction may be performed using the vertical projection pattern of each character line instead of directly scanning the buffer memory.

Furthermore, if there is no continuous character line in the first line, it goes without saying that it is necessary to detect and extract characters from the continuous character line.

〔Effect of the invention〕

As described above, according to the present invention, fluctuations in the conveyance speed of documents,
Even if there is a change in the position of printed or handwritten characters on a form, the characters can be cut out and recognized with very little effort. It is particularly effective for cutting out and recognizing blank spaces (nupaine).

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a character recognition device according to an embodiment of the present invention, FIG. 2 is a pattern diagram showing a specific example of a reading pattern of the device, and FIG. 8 is a flowchart showing the operation of the device. 1... Photoelectric conversion unit, 2... Buffer memory, 8...
-Character recognition unit, 4...control unit, 5...external control device. Applicant's agent Kiyoshi Inomata

Claims (1)

[Claims] A reading unit that reads a form in which at least one character line is a continuous character line with no spaces in between, and a reading pattern of the form read by the reading unit. a storage section for cutting out the pattern of each character line of the reading pattern stored in the storage section, a cutting section for cutting out the pattern of each character line of the reading pattern stored in the storage section, one character at a time, and a recognition section for recognizing the character pattern cut out by the cutting section. In the character recognition device equipped with the above, the cutting unit cuts out each character pattern from the continuous character line in the reading pattern, and calculates an average positional error between this cutting position and a predetermined temporary cutting position;
A character recognition device characterized in that a pattern of Noh character lines is cut out by correcting a predetermined temporary cutting position 1f for Noh character lines using the average position error.