JPS62281082A - Character recognizing device - Google Patents

Abstract

PURPOSE:To attain the corrector character recognition by outputting the evaluation based upon the difference of the similarity of an input character together with a recognizing candidate by matching with a recognizing result. CONSTITUTION:At a matching part 4, a large classification is executed, the characteristic outputted from a characteristic extracting part 3 is matched with some of the standard patterns stored in a dictionary part 5, and plural sets of the difference or similarity with the information of the character type of a standard pattern which comes to be the object of the matching are set to a character candidate deciding part 6. Next, when it is judged that the mode set earlier is not the estimation output mode, the character candidate deciding part 6 executes the output as the recognizing result based upon the difference of similarity out of the results of plural matchings, selects the candidate of a character code and sends it to a result output part 8. When the candidate cannot be selected, the operation is returned to a pre-pressing part 2 and re- executed, and when re-execution exceeds the constant number of times, the non-recognizable signal is sent to the result output part. When it is judged that the mode is the evaluation output mode, the non-recognizable signal is sent to an evaluation deciding part 9.

Description

Detailed Description of the Invention 3. Detailed Description of the Invention [Field of Industrial Application] The present invention relates to a character recognition device that recognizes input character symbols.

[Prior Art] Conventionally, this type of device has been configured to output only character codes that are candidates for the input character as a recognition result, or a signal indicating that the input character is unrecognizable. There were some warm points regarding the type of writing style that would allow for more accurate recognition.

In addition, even if writing samples are posted, character recognition devices differ in the writing style and shape of input characters that can be recognized more accurately depending on the type and recognition method. In some cases, you may not get the correct result even if you enter the characters you intended to write, and in some cases, characters written in a different way than the sample may be recognized correctly.

[Problems to be Solved by the Invention] The present invention has been made in view of the above-mentioned prior art, and is capable of outputting an evaluation based on the degree of difference or similarity between an input character and a standard pattern together with a recognition result. Another object of the present invention is to provide a character recognition device that reduces the conventional warmth described above.

[Means for solving the problem] In order to solve this problem, the present invention has the following configuration.

That is, an input means for inputting character symbols, a recognition means for recognizing the character symbols input by the input means and outputting recognition candidates, and a recognition means for inputting the characters input by the input means with respect to the candidates recognized by the recognition means. and notifying means for notifying the degree of similarity or difference of character symbols.

[Operation] In the configuration of the present invention, the recognition means recognizes the character symbol inputted by the input means and outputs the recognition candidate, and the degree of similarity or dissimilarity between the recognition candidate of the input character symbol is notified by the notification means. inform.

[Examples] Examples according to the present invention will be described in detail below with reference to the accompanying drawings.

FIG. 1 is a block diagram of a character recognition device in this embodiment.

In the figure, 1 is a character input unit, which is a scanner or online in the case of an optical character recognition device, and a digitizer in the case of handwritten character recognition. A preprocessing unit 2 binarizes input image data if it is multivalued, and performs necessary preprocessing such as noise removal, smoothing, line thinning, etc.

Reference numeral 3 denotes a feature extraction section which extracts features necessary for matching from the image data processed by the preprocessing section 2.

As the feature, one or a combination of outlines, backgrounds, core lines, etc., is used. Reference numeral 4 denotes a matching section, which performs matching between the characteristics of the input characters and standard patterns in the dictionary section 5, which will be described later, to calculate the degree of difference or similarity. The matching section 4 also performs the work of preliminarily narrowing down the standard patterns to be matched (major classification, etc.). 5 is a dictionary section which stores standard patterns. A character candidate determination unit 6 selects character code candidates to be output as recognition results based on the character type of the standard pattern to be matched and the degree of difference or similarity. Reference numeral 7 denotes a mode signal input section which sends a signal to the character candidate determination section 6 indicating whether or not to output an evaluation of the input character based on the degree of difference or similarity between the input character and the standard pattern. still,
This mode signal input section 7 may be energized by a switch provided outside the device, or may be controlled by an external device. 8 is a result output unit, which may be a printer, a CRT, or an interface to an electronic computer. Reference numeral 9 denotes an evaluation determination unit that evaluates the input character based on the degree of difference or similarity between the input character and the standard pattern, and ranks the input character in 4 stages of AND, for example. The IQO is a CPU'U that controls the entire device, and operates according to the processing procedure of the flowchart in FIG. 2, which will be described later. Note that the program corresponding to this flowchart is stored in the ROM 101. 102 is a RAM used as a work area for the CPU 100.

An example of this is shown in FIGS. 3(a) and 3(b).

Figure 3 (a) shows the recognition results and evaluations when Operator 1 Nota inputs the letter ``A'' consciously, and in both cases, the recognition result is ``A''. .

In the case of the input character 30, the line segments (vectors) of the three trees forming the character r A J are almost the same as the standard pattern "A", so the evaluation is A°. In the case of the input character 31, the line segment which is the horizontal line ゛°-°° in the character "A" is placed within a predetermined width, but since this line segment is tilted, the evaluation is B'.

It becomes. The horizontal line of the input character 31 protrudes from the input character 32. Normally, in recognition processing, this protruding line may be determined to be another line segment, so the recognition rate for the character "A" as a recognition result is not very good.

Now, for input character 33, each of the three line segments has a "jump" or "jumping line segment", and the recognition result is the character "A", but no other recognition candidates were found. This indicates a state in which there was no recognition result, and a state in which the recognition result has almost no reliability.

As described above, the evaluation of the input characters and symbols is output, but the criteria for recognition determination described above differ depending on the recognition method, so the evaluation is not limited to this recognition method. Furthermore, the third
Figure (b) shows an example in which recognition results and recognition evaluations for other input characters are output.

The operation of the character recognition device of this embodiment having the above-mentioned configuration will be explained with reference to the flowchart shown in FIG.

First, in step S1, a mode is set. This mode determines whether or not to output the similarity or dissimilarity of the character symbol input from the input unit 1 with respect to the recognition candidate output as the recognition result. Next, step S2'
Image data of characters and symbols is inputted from the C input section 1, preprocessed by the preprocessing section 2, and sent to the feature extraction section 3, whereupon the process proceeds to recognition processing. In step S3, features are extracted from the image data sent to the feature extraction section 3, and the extracted feature information is sent to the matching section 4. The matching unit 4 performs broad classification, matches the features output from the feature extraction unit 3 with some of the standard patterns stored in the dictionary unit 5, and compares the character type information of the standard patterns targeted for matching. A plurality of sets of degrees or similarities are sent to the character candidate determination section 6. Next, in step S4, it is determined whether the previously set mode is the evaluation output mode. If it is determined that it is not the evaluation output mode in this determination, the process moves to step S4, and the character candidate determination unit 6 outputs the recognition result based on the degree of dissimilarity or similarity from among the plurality of matching results, and outputs the character code as a recognition result. The candidates are selected and sent to the result output unit 8. If no candidate can be selected, the process returns to the preprocessing unit 2 and tries again, but if the process is repeated a certain number of times, the process does not start again and sends an unrecognizable signal to the result output unit 8.

Further, if it is determined in step S4 that the mode is the evaluation output mode (if a signal for outputting the evaluation of the input character is sent from the mode signal input section 7 to the character candidate determination section 6), the character candidate determination section 6 sends the character code or an unrecognizable signal to the result output unit 8, and also sends the degree of difference or similarity corresponding to the character type of the character code or an unrecognizable signal to the evaluation determination unit 9. The evaluation determining section 9 performs evaluation based on the output from the character candidate determining section 6, and sends the evaluation result to the result output section 8.

As explained above, according to the present embodiment, an evaluation based on the degree of difference or similarity between input characters and recognition candidates is output together with the recognition results, which is useful for educating scribes regarding character input. to make students aware of accurate writing methods,
More accurate character recognition becomes possible.

Further, in this embodiment, the recognition results are evaluated in four stages A to D, but they may be output as a percentage, for example.

[Effects of the Invention] As explained above, according to the present invention, an evaluation based on the degree of difference or similarity between an input character and a recognition candidate is output together with a recognition result, which is useful for training scribes on character input. It is useful to make scribes aware of handwriting methods that can be recognized accurately, and enables more accurate character recognition.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of the character recognition device of this embodiment, Fig. 2 is a flowchart for explaining the operation of the character recognition device of this embodiment, and Fig. 3 (a) and (b) are this embodiment. FIG. 3 is a diagram showing recognition results and recognition evaluations for input characters in FIG. In the figure, 1...character input section, 2...preprocessing section, 3...
・Feature extraction unit, 4...Matching unit, 5...Dictionary unit, 6...Character candidate determination unit, 7...Mode signal input unit, 8...Result output unit, 9...Evaluation Judgment section, 30-3
3... Input characters, 100・CPU, 101・ROM,
102...RAM. Patent applicant: Canon Co., Ltd. -He43- (b) Figure 3

Claims (4)

[Claims] (1) An input means for inputting character symbols, a recognition means for recognizing the character symbols input by the input means and outputting recognition candidates, and an input means for inputting the candidates recognized by the recognition means using the input means. 1. A character recognition device comprising: notification means for notifying the degree of similarity or difference between character symbols. (2) The character recognition device according to claim 1, wherein the notifying means notifies the degree of similarity or difference in predetermined stages. (3) The character input device according to claim 1, wherein the input means is handwritten input. (4) The character recognition device according to claim 1, wherein the input means optically reads character symbols recorded on the original.