JPH0355679A - Collating method between character recognition result and character input result - Google Patents

Abstract

PURPOSE:To improve the accuracy and the reliability of the input result characters by selecting one or more candidate characters based on the resemblance to compare them with the input result characters and deciding the coincidence between both characters. CONSTITUTION:A candidate recognition result character Xi is read out of a memory 4 to a subject character X. At the same time, an input result character X0 is read out of a memory 5 to the character X. A counter (i) related to the character Xi is initialized to '1'. Then the presence or absence of the character Xi is decided. The coincidence, if obtained between both characters Xi and X0, is displayed on a CRT. If no coincidence is obtained between both characters, the due process is repeated. In such a constitution, the useless processes can be omitted for the collation and the correction caused by the discordance between both characters. Then the collation/correction processing efficiency is improved.

Description

[Detailed description of the invention] [Industrial application field]

This invention provides a method for collating character recognition results and character input results;
That is, it relates to a method of comparing a recognition result character when a target character to be recognized is recognized via a character recognition device with an input result character that is an output when the target character is input via a keyboard.

[Conventional technology]

BACKGROUND ART Conventionally, when inputting a large amount of characters for a document in order to encode the document, the following methods have been used to improve the reliability. First, a document is converted into an image using an image scanner, and the converted output is recognized as characters by a character recognition device to create matching data. This verification data is verified by a program against the verification data input by the operator. If there is a discrepancy between the two, it is determined which error was caused by that location, and based on this determination, this location is corrected by key personnel until the error is eliminated.

[Problem to be solved by the invention]

As explained above, in the conventional technology, even if the verification data by the operator is correct, if there is an error in the verification data by the character recognition device, a discrepancy is displayed, and it is difficult to determine which is the correct difference. required. Traditionally, these discrepancies often occur due to errors in character recognition devices, often requiring unnecessary judgment.
As a result, the efficiency of verification work was significantly hindered. An object of the present invention is to solve the above-mentioned problems of the conventional technology, and to provide a recognition result character when a target character to be recognized is recognized through a character recognition device, and a recognition result character when the target character is inputted via a keyboard. It is an object of the present invention to provide a method for collating character recognition results and character input results, which more efficiently collates the input result characters that are output when a character is recognized.

[Means to solve the problem]

In order to solve this problem, a method for matching character recognition results and character input results according to the present invention provides a recognition result character when a target character to be recognized is recognized via a character recognition device,
In the method of comparing the target character with a human result character that is an output when the target character is input via a keyboard, selecting one or more candidate characters as the recognition result character based on the degree of similarity: the input result character is the same as any of the candidate characters of the recognition result characters, determining that the matching result is a match;

[For use]

One or more candidate characters are selected as recognition result characters based on similarity. Each candidate character is then compared with the human result character, and if the input result character is the same as any of the candidate characters, the match result is determined to be a match, and if it is not the same as any of the candidate characters, the match is determined to be a match. The result is determined to be a mismatch. Therefore, by determining that there is a mismatch, it is determined that there is an abnormality in at least one of the corresponding input result character or the candidate character as the recognition result character, and this can be corrected.

【Example】

A collation device to which a method for collating character recognition results and character input results according to the present invention is applied will be described below with reference to the drawings. FIG. 2 is a block diagram showing the structure of this verification device. In the figure, 1 is an image scanner for the target document;
is a recognition unit that performs character recognition based on the output of the image scanner 1. Note that, unlike conventional methods in which only one character with the highest degree of similarity to the target character is output as a recognition result, this recognition unit 2 outputs the character in the order of the highest degree of similarity or by relaxing the threshold related to the degree of similarity. In any case, up to three characters are output as candidate characters. Reference numeral 3 denotes a keyboard on which characters are input by the operator while viewing the target document. 4 is a memory for data related to the recognition result characters from the recognition unit 2; Further, 5 is a memory for data related to input result characters from the keyboard 4. 6 is a matching unit that matches the corresponding character data from each memory 4 and 5 to find a match. Determine inconsistency. 7 is CR
Press T to display the matching results on the screen. It is also possible to install a printer that prints and outputs the verification results together with the CRT 7. Now, FIG. 3 is a diagram showing the structure of a data file of recognition result characters in the memory 4. In the same figure, each address 0
01,002,003 XI. X2. X3 is a candidate character for the recognition result for the target character X, none of which is shown in the figure, and in this case, up to three characters are selected, which is the limit. Further, YLY2 at each address 004.005 is a candidate character for the recognition result for the target character y (not shown), and in this case, only two characters were selected. Similarly, Zl at address 006 is the only candidate character for the recognition result for target character 2 (not shown). The operation of the collation device will be described with reference mainly to the flowchart in FIG. 1 and supplementary to the block diagram in FIG. 2. In FIG. 1, in step S1, a candidate character Xi of the recognition result character for the target character X is read out from the memory 4 (see FIG. 2), and an input result character Xo for the target character X is read out from the memory 5. In step S2, a counter i related to candidate character Xi is initialized, i=1. In step S3, the presence or absence of Xi is determined, and if YES (present), the process moves to step S4, and if No (absent), the process moves to step S7, where the verification result is NG. In step S4, it is determined whether the candidate character Xi matches the input result character XO, and YES (match) is determined.
If so, the verification result becomes OK in step S8. No(
If not, the process goes through steps S5 and S6, and then goes to step S.
The procedures from 3 onwards are repeated for all candidate characters. When the repetition is completed, the process moves to step S7 and ends. After steps S7 and 3B, the matching result is displayed on the CRT in step S9, and the matching process for the target character X is completed. Subsequently, matching for the next target character will be performed in the same way.

【Effect of the invention】

As explained above, the present invention has the following superior effects compared to the conventional technology because it compares one or more candidate characters as recognition result characters with input result characters. In other words, it is possible to improve the reliability of the accuracy of characters recognized and input based on the match judgment, and
It is possible to prevent wasted verification processing and subsequent correction processing due to mismatch judgments that occur frequently and needlessly, and to improve verification efficiency. Correction processing can be made more efficient.

[Brief explanation of drawings]

FIG. 1 is a flowchart showing the verification operation of a verification device to which the method of the present invention is applied, FIG. 2 is a block diagram showing the configuration of this verification device, and FIG. 3 is the structure of a data file of recognition result characters in this verification device. This is a diagram. Code explanation 1: Image scanner, 2: Recognition unit,

Claims (1)

[Claims] 1) A method of comparing a recognition result character when a target character to be recognized is recognized via a character recognition device with an input result character that is an output when the target character is input via a keyboard, selecting one or more candidate characters as the recognition result character based on similarity; determining that the collation result is a match when the input result character is the same as any of the candidate characters of the recognition result characters; A method for matching a character recognition result and a character input result, comprising the following steps.