JPS6318483A - Character recognizing method for optical information input device - Google Patents

Abstract

PURPOSE:To surely recognize a character string connecting characters by segmenting a proposed character pattern position from input image information, to execute pattern matching, and at the time of a reject, determining a proposed extracting position from the recognized result or the like obtained just before the rejection. CONSTITUTION:A character is segmented from a character string optically read out by a scanner 11 in an optical input device 2 at a proposed position by a character extracting unit 12 and a feature pattern 12 is extracted by a feature extracting unit 13 and matched with a dictionary pattern stored in a feature dictionary memory 15. At the time of reject, the recognized result obtained immediately before the proposed position is referred and plural proposed extracting positions are forward moved by one position to execute matching. When plural proposed position do not exist, the recognized result obtained immediately before is referred, and at the time of existence of a recognized result, an extracting point of the just preceding character pattern is forward moved by one point. At the time of no existence of a recognized result, the segmented recognition is decided as a reject. Thus, characters connected to each other in a character string can be surely recognized by the segmenting matching.

Description

[Detailed Description of the Invention] [Technical Field] The present invention relates to a character recognition method in an optical input device that optically reads and inputs information on a document, and in particular, to a method for reliably recognizing character strings in which single characters are connected to each other. It is related to the techniques used to perform this task.

[Prior Art] An input device for an image information processing apparatus using a computer, such as an optical input bag that optically reads and inputs information on a document! i! (hereinafter referred to as OCR), generally,
Input characters (input image information) are recognized using a recognition method based on template (dictionary memory) matching method.

In the recognition method based on this template matching method, the template corresponding to each character is

As shown in FIG. 4, a character pattern with a width W or a pattern corresponding thereto is provided as a standard pattern (dictionary).

Then, characters are separated and recognized from the input image information using a character extraction method using vertical projection.

However, in the conventional character extraction method using vertical projection, when a pattern sequence in which it is difficult to separate one character is given as shown in FIG. There are too many C1□, so the V-work shown in Figure 5,
V2. V3... Because it will be disconnected incorrectly,
This was a cause of erroneous recognition and jets.

[Objective] The object of the present invention is to provide a technology that can reliably recognize a printed character string in which single characters are connected to each other in a character recognition method using an optical input device that optically reads and inputs information on a manuscript. It is about providing.

The above and other objects and novel features of the present invention will be explained by the following description of the specification and the accompanying drawings.

[Structure] The present invention provides a character recognition method for an optical input device that optically reads and inputs information on a document, including a step of extracting an image block of interest from input image information, and a step of extracting the extracted image block of interest. a step of determining a candidate point for a character pattern cutting position from Referring to the recognition result immediately before the cutting position of the candidate point, if there are multiple cutting position candidate points immediately before the cutting position of the candidate point, move the cutting point of the current character pattern forward by one, and move this to the current character pattern. a cutting position candidate point is cut as a cutting position candidate point, and character recognition is performed by checking the matching of the character pattern at the cutting position with a character pattern in a dictionary; If not, refer to the recognition result immediately before the current image block of interest, and if there is a recognition result, move the cutting point of the character pattern immediately before the current character pattern one step forward, and perform recognition of that character pattern. The present invention is characterized by comprising the step of determining that character pattern cutout recognition is rejected when there is no result.

[Example] An example of the present invention will be specifically described below with reference to the drawings.

In addition, in all the times for explaining the example.

Components having the same function are given the same reference numerals, and repeated explanations thereof will be omitted.

FIG. 1 is a block diagram showing a schematic configuration of an OCR input device according to a character recognition method for a CR input device according to an embodiment of the present invention, and FIG. A block diagram showing a schematic configuration of a functional image information processing device. FIG. 3 is a flowchart of a character recognition method for an OCR input device according to an embodiment of the present invention.

As shown in FIG. 2, the image information processing apparatus to which the character i1 recognition method for an OCR input device of this embodiment is applied has a keyboard 1 for inputting a single character as well as various modes (kana-kanji conversion, kanji-kana conversion, etc.). , OCR character recognition, etc.). The OCR input device 2 optically reads and inputs a document. The processing device 3 executes processing according to a designated mode on the input information from the keyboard 1 and the OCR input device 2, and outputs the processed information to the output device 4. Output device i!
4 is a general term for display devices, printers, etc. It is stored in the program memory (ROM) 5 necessary for the processing of the processing unit 8, but it is possible to convert kana to kanji by keyboard input, OCR characters! ! ! A common algorithm is used as much as possible for the post-processing of information, the conversion of characters input by OCR into kana/kanji, and kanji/kana. Data memory (RAM) 6 is.

Processing device! ! It is used to store data and parameters that are being processed in 3. The word dictionary memory 7 stores a word dictionary to which reading orthography correspondence data is added.

As shown in FIG. 1, the OCR input device 2 shown in FIG. 2 reads and inputs image information such as characters on a document using an optical scanner 11 comprising a light source and a charge-coupled device (COD). . The image information such as the inputted kana character string, kana-kanji mixed character string, alphabetic character string, etc. is processed by the character cutting unit 12 into one character (width W) as shown in FIG.
□ to W S ), and the feature extraction unit 13 extracts the features of the cut out characters. This extracted data is matched with feature dictionary data stored in a feature dictionary memory (ROM or RAM) 15 in a feature matching unit 14. If matching is achieved, the input characters are recognized and sent to the processing device 3. If the matching is rejected, a reject signal is generated from the feature matching unit 14 and sent to the character cutting unit 12.

Next, the processing process of the character recognition method for an OCR input device of this embodiment will be explained according to the flowchart shown in FIG.

First, in step 101, an image block of interest is extracted as shown in FIG. This extracted image block of interest is an image block including at least several characters, and its length may be fixed length or variable length (word).

Next, in step 102, the check pack trace (Bac
k Traca) Set the counter that sets the code to 0. This means that pack tracing (backtracking) is not performed to extract one character.

Next, in step 103, cropping position candidate points for the image block are determined. That is, for example, in the case of FIG. 6 corresponding to the character string pattern shown in FIG. 5, the cutting point C7, which is a candidate cutting position point, is determined.

Next, in step 104, it is determined whether or not there is a candidate cropping position point for the image block, and if there is a candidate cropping position point (cutting point c,) for the image block as in the example of FIG.
No), proceed to step 105, and if there are no candidate points (YES).
), move to stage W1112. In step 105, character pattern cutout is recognized.

That is, recognition is attempted by regarding the cutting points 01 to C1 in FIG. 6 as one character.

Next, recognition was attempted in step 105, but that! ! ! It is judged in step 106 whether or not the recognition has been achieved (matching has been achieved). If the recognition is possible (YES), the process moves to step 107. If the recognition is not possible (NO), the process moves to step 108. In the example shown in FIG. 6, in which the character pattern cutting position candidate point is moved forward one position and returned to step 103, the cutting point C7 is changed to the previous cutting point C1, and steps 104 and 106 similar to the above are performed. try. Furthermore, if the character cannot be recognized, the first step 108 is repeated and the cutting points are changed to C, C4 shown in FIG. 6, at which point the character can finally be recognized.

Up to this point, recognition of all the characters in the image block has not yet been completed, so cutting points 04 to C1 are set as candidate cutting points using the same criteria as above.

That is, in step 107, the recognition result (character code string) of backtrace=O is output. Based on the output data of this recognition result, it is determined in step 109 whether or not character recognition in the image block of interest has been completed. If it has been completed (YES), the process moves to step 110 and the block extraction range is moved.

Next, in step 111, it is determined whether there is any image block to be extracted, and if there is no image block to be extracted (YES), O
The processing process of the character recognition method for CR input device is completed. If there is an image block to be extracted (No), the process returns to step 101.

Next, in the step 104, it is determined whether there is a candidate cropping position point for the image block, and if there is no candidate cropping position point for the image block (YES), the process moves to step 112. That is, in the example shown in FIG. 6, the primary target character pattern is the cutting points C1° to C, but this is because there is no corresponding character! ! Really fail. Therefore, the cutting point C1°~C
Since there is no candidate point to cut between 01 and 01, it is checked in step 112 whether the back trace counter is ``1'', and ``
If the back trace counter is "1" (YES), proceed to step 116, perform a reject process, and return to step 105; if the back trace counter is "0" (NO), step 113
Then, it is determined whether there is a recognition result immediately before the current image block, and if there is a recognition result (YES), the process moves to step 114. If there is no recognition result (No), the process moves to step 116, a reject process is performed, and the process returns to step 105.

In step 114, the cutting point of the pattern of the immediately previous image block is moved forward by one, and in step 115, it is determined whether there is a cutting candidate. If there is a cutting candidate (No), step 1 is performed.
Return to 05, if there are no cutting candidates (YES), step 11
6, a reject process is performed and the process returns to step 105.

That is, in the example shown in FIG. 6, in step 114, the cutting point C8° of the pattern of the immediately previous image block is moved to the previous cutting point Cs, and the back trace counter is set to "1". The pattern of cutting points 07 to C9 is recognized and one character "u" is recognized. At this point, the back trace counter becomes "0".

Next, a pattern with cutting points Cg to C1□ is cut out.

When the same processing as above is performed, one character rnJ is recognized. Finally, the cutting points 011 to C1□ are recognized, and all processing of the block of interest is completed.

The present invention has been specifically explained above using examples.
1. It goes without saying that the present invention is not limited to the embodiments described above, and can be modified in various ways without departing from the spirit thereof.

〔Effect of the invention〕

As described above, according to the present invention, the following effects can be obtained.

In a character recognition method using an optical input device that optically reads and inputs information on a document, there are two steps: extracting an image block of interest from input image information, and determining the cutout position of a character pattern from this extracted image block of interest. A step of determining a candidate point, a step of checking the matching between the character pattern cut at this candidate point and the literary pattern in the dictionary, and if the result of this matching is a reject,
Refer to the recognition result immediately before the cutting position of the candidate point.

If there are multiple cutting position candidate points immediately before the cutting position of the candidate point, move the cutting point of the current character pattern forward by one, cut this as the cutting position candidate point of the current character pattern, and set the cutting position. character recognition by looking at the matching between the character pattern and the character pattern in the dictionary; and if there is no cutting position candidate point immediately before the cutting position of the cutting position candidate point, the recognition result immediately before the current image block of interest; If there is a recognition result, move the cutout point of the character pattern immediately before the current character pattern one step forward, and if there is no recognition result for that character pattern, character pattern cutout 1m is rejected. Therefore, it is possible to reliably recognize all the characters in the image block of interest, which is composed of character strings such as printed characters in which the extracted characters are connected to each other.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a schematic configuration of an OCR input device according to a character recognition method for an OCR input device according to an embodiment of the present invention. FIG. FIG. 3 is a flowchart of a character recognition method for an OCR input device according to an embodiment of the present invention. FIG. 3 is a diagram for explaining character extraction processing in a character recognition method for an input device. In the figure, 3...processing device, 11...scanner, 12
...Character extraction unit, 13...Feature extraction unit. 14... Feature matching unit, 15... Feature dictionary memory.

Claims (1)

[Claims] (1) In a character recognition method using an optical input device that optically reads and inputs information on a document, there are two steps: extracting an image block of interest from input image information, and extracting a character pattern from the extracted image block of interest. a step of determining a candidate point for the cutting position, a step of checking the matching of the character pattern for cutting at this candidate point with a character pattern in the dictionary, and, if the result of this matching is a reject, a step of determining the cutting position of the candidate point. Referring to the previous recognition result, if there are multiple cutting position candidate points immediately before the cutting position of the candidate point, move the cutting position candidate point of the current character pattern forward by one, and use this as the cutting position candidate point of the current character pattern. A step of cutting as a position candidate point and recognizing a character by checking the matching between the character pattern at the cutting position and a character pattern in a dictionary, and when there is no cutting position candidate point immediately before the cutting position of the cutting position candidate point. , refers to the recognition result immediately before the current image block of interest, and if there is a recognition result, moves the cutting point of the character pattern immediately before the current character pattern to the previous candidate point, and calculates the recognition result of that character pattern. 1. A character recognition method for an optical information input device, comprising the step of rejecting character pattern cutout recognition if there is no character pattern.