JP3221968B2 - Character recognition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an easy-to-operate character recognition device capable of processing a form or document quickly and accurately.

[0002]

2. Description of the Related Art Conventionally, knowledge dictionaries have been used to improve the recognition rate of handwritten characters. The information held by the knowledge dictionary includes information on words to be recognized, context information,
And other information. The user of the character recognition device can newly register such information. Less than,
As a conventional character recognition technology using the word information of the knowledge dictionary, for example, a technology disclosed in Reference 1: A Collection of 1982 IEEJ General National Convention Lectures, Volume 5-326 will be described.

First, a predetermined area of a form or document is optically scanned, and an optical signal from a sheet is photoelectrically converted to obtain image data of the form or document. Then, a pattern of a character to be recognized is cut out from the image data. Characters to be recognized are recognized based on the extracted character pattern, and one or more candidate characters are obtained as a recognition result. Then, for one unit of knowledge processing characters, one or more candidate characters obtained for each character are combined to form a character string, and the candidate rank or similarity assigned to each candidate character in the character string is used. Then, the similarity between the character string and the word of the word information is calculated, and the character string having the maximum similarity is selected. Then, the word of the word information corresponding to this character string is displayed. However, when the similarity of the character string is equal to or smaller than a predetermined threshold, a character string formed by combining candidate characters having the first candidate rank is displayed.

When correcting the knowledge processing result, a word group of word information whose similarity with the character string is equal to or more than a predetermined threshold is displayed as a candidate word, and a correct word is included in the word group. Correction is made by selecting it.

[0005]

In the above character recognition technology, when displaying a candidate word, (1) a case where the word of the word information that originally existed is output as a candidate word;
(2) It is conceivable that the word of the word information registered by the user is output as a candidate word, but the user registers word information having high necessity according to the purpose (2)
The word in case (1) is more likely to be correct than the word in case (1). Nevertheless, the display of words in cases (1) and (2) is the same. Therefore, since the operator cannot find a word that is likely to be correct only by glance at the display of candidate words, it is necessary to carefully observe all displayed words and select a correct word, For this reason, there has been a problem that editing work such as correction and confirmation requires much time, and the number of forms or documents that can be processed per unit time is reduced.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a character recognition device which solves the above-mentioned conventional problems and makes it easier to select a correct word from displayed candidate words, thereby improving operability. Is to do.

[0007]

According to the present invention, there is provided a recognition section for outputting a recognition result of a character pattern cut out from image data of a quantized form or document, and a word dictionary for storing word information. A word registering unit for registering the word, a post-processing unit that outputs a knowledge processing result and a candidate word based on the recognition result using the knowledge dictionary, a result editing unit that edits the post-processing result, and the candidate word And a display unit for displaying a word, and an input unit for inputting a correct word, the character recognition device, the display unit, when the candidate word is a word of the word information originally existed in the knowledge dictionary, It is characterized in that the display method of the word is changed depending on whether the word is a word of the word information registered in the word registration unit.

[0008]

According to the present invention, the display of the display unit when selecting the correct answer from the candidate words is performed when the word of the word information existing from the beginning in the knowledge dictionary is output as the candidate word and when the word is registered in the word registration unit. When the word of the word information is output as a candidate word, the word of the word information existing in the knowledge dictionary from the beginning in the operation of selecting the candidate word and the word of the word information registered by the user are changed. Since the operator can easily identify the candidate word, the operator can instantly find a candidate word having a high possibility of correct answer, and perform operations such as correction and confirmation. Therefore, the above problem is solved.

[0009]

Embodiments of the present invention will be described below with reference to the drawings. It should be noted that the drawings are only schematically shown to an extent that the present invention can be understood.
The arrangement positions, dimensions, input / output signals, and connection relationships are not limited to the illustrated examples.

FIG. 1 is a functional block diagram for explaining one embodiment of the present invention. Character recognition device 10 of this embodiment
Is a recognition unit 12 that cuts out a character pattern from image data of a quantized form or document, and outputs a recognition result of the cut out character pattern, and a word registration that registers word information desired by a user of the character recognition device. Unit 14, a post-processing unit 16 that outputs knowledge processing results and candidate words of recognition results based on the knowledge dictionary, a result editing unit 18 that corrects and confirms the knowledge processing results, and a display that displays the knowledge processing results and candidate words. Unit 20, an input unit 22 for inputting correct words, and further includes a word registration unit 14, a result editing unit 18, and a display unit 2.
0 and a control unit 24 for controlling the operation of the input unit 22. In FIG. 1, reference numeral 26 denotes a photoelectric conversion unit that outputs quantized image data of a form or document.
Reference numeral denotes an image memory for storing image data from the photoelectric conversion unit 26.

FIG. 2 is a diagram showing an example of a form, in which 30 is an example of a form in which an address is described, and 32 is an entry frame for designating a character description area.

FIG. 3 is a diagram showing an example of selection of candidate words. In FIG. 3, reference numeral 34 denotes a display screen, 36 denotes a form, 38 denotes a character string area to be corrected, 40 denotes a cursor position, 42 denotes a candidate word display frame, Reference numeral 44 denotes a user registration word whose display method has changed.

Hereinafter, the present embodiment will be described in more detail with reference to FIGS. 1, 2, and 3. FIG. The photoelectric conversion unit 26 optically scans a predetermined reading range on a form or document,
The optical signal L from the form or document is photoelectrically converted to output image data quantized to black and white binary, and the image memory 28 stores the image data.

The recognizing unit 12 extracts a character pattern from the image data in the image memory 28, and extracts various features relating to a character to be recognized from the extracted character pattern.
Then, the feature of the extracted character pattern is compared with the feature of the standard character pattern, and the character recognition result and the candidate ranking are output.
One or more candidate characters are obtained as a recognition result for one character. If there is only one candidate character, candidate rank 1 is added to the candidate character and output. If there is more than one candidate character, Outputs the candidate rank determined for each candidate character to the candidate character.

The word registration unit 14 additionally registers word information desired by the user of the character recognition device in the knowledge dictionary. The word information of the knowledge dictionary is added with information that can distinguish the word information registered by the word registration unit from general word information prepared from the beginning. Hereinafter, this information is referred to as word registration information.

The post-processing unit 16 performs knowledge processing using word information based on the recognition result from the recognition unit 12. Post-processing unit 1
When a recognition result of a character of a knowledge processing one unit sentence (for example, a recognition result of a description area of a prefecture name in the form 30 shown in FIG. 2) is input, candidate characters of each character for one unit of the knowledge processing can be combined. The character string is collated with the word of the word information, and it is checked whether a word corresponding to the character string including the candidate character exists in the word information. Then, when a character string A that matches the word of the word information is detected from the character strings formed in combination, the evaluation value J of the character string A is calculated. Where S
Is the sum of the candidate ranks assigned to each candidate character of the character string and N represents the total number of characters constituting the character string, and the evaluation value J can be expressed as, for example, J = S ÷ N. .

The determination as to whether or not the word and the character string A match is performed, for example, based on whether or not all the character codes of the characters at the corresponding positions of the word and the character string A match. Then, when all the character strings formed for one unit of the knowledge processing have been collated with the word information, the character string A with the smallest evaluation value J is selected from the character strings A as the knowledge processing result. Further, word information having a predetermined evaluation value or less is selected as a candidate word of the character string A. This character string A is sent to the result editing unit 18 together with the candidate words.

Further, if only one character string A is detected when all the character strings of the knowledge processing one unit sentence have been collated with the words of the word information, the character string A is selected as a knowledge processing result and selected. The resulting character string A is sent to the result editing unit 18.

If all character strings for one unit of knowledge processing have not been detected when all the character strings for one unit of knowledge processing have been collated with the words of the word information, the candidate rank of each character for one unit of knowledge processing is changed. A character string A formed by combining the first-place candidate character is selected as a knowledge processing result, and the selected character string A is sent to the result editing unit 18.

If there is a candidate word for the character string A sent from the post-processing unit 16, the result editing unit 18 sends the character string A and the candidate word to the control unit 24. And the control unit 24
Sends a display instruction of the character string A, the candidate word, and the candidate word according to the word registration information to the display unit 20. At this time, if the candidate word is not the word of the word information registered in the word registration unit, the instruction to display in the first color is given. If the candidate word is the word of the word information registered in the word registration unit, the first color is displayed. And an instruction to display in a second color different from. Further, the display unit 20 displays the character string A and the candidate words by the method specified by the display instruction (for example, the display screen 34 shown in FIG. 3). Here, the operator determines whether or not the character string A in the display screen is erroneous, and if it is erroneous, selects the correct word from the candidate words. If there is no correct word in the candidate words, the correct word is input using a keyboard or the like.

When there is no candidate word for the character string A, the result editing unit 18 sends the character string A to the display unit 20 via the control unit 24, and the display unit 20 displays the character string A. Here, the operator determines whether or not the character string A in the display screen is incorrect, and if so, inputs the correct word using a keyboard or the like. Then, the result editing unit 18 receives the above-described correct word via the input unit 22 and the control unit 24, and replaces the correct word with an incorrect character string. When the character string A displayed on the display unit 20 is determined to be correct by the operator, the result editing unit 18 outputs the character string A as it is.

The present invention is not limited only to the above-described embodiment, and accordingly, the configuration, operation, processing contents, input / output signals and numerical conditions of each component may be arbitrarily and suitably changed. For example, in the above-described embodiment, the user additionally registers word information in the general knowledge dictionary in the word registration unit, but prepares another user-specific knowledge dictionary and registers necessary word information there. Is also good.

Further, in the above-described embodiment, the value obtained by dividing the sum S of the candidate ranks assigned to each candidate character of the character string by the total number N of characters constituting the character string is used as the evaluation value J.
The score corresponding to each candidate rank in place of the sum S of the candidate ranks (for example, a score that becomes lower as the candidate rank decreases, such as associating 100 points for candidate rank 1 and 90 points for candidate rank 2). May be used. Alternatively, the sum of the frequency of appearance of each candidate character of the character string (in this case, the frequency of appearance is held in advance by the recognition unit) may be used instead of the sum S of the candidate ranks.
Alternatively, a similarity between a candidate character and a character pattern corresponding to the candidate character may be determined in place of the candidate sum S, and the sum of the similarities of the candidate characters in the character string may be used. Alternatively, instead of the sum S of the candidate characters, the distance between the dictionary matrix of the candidate characters and the characteristic amount of the character pattern corresponding to the candidate characters is obtained, and the sum of the distances of the candidate characters of the character string may be used. Good. Alternatively, instead of the sum S of the candidate ranks, the sum of the appearance frequencies of the candidate characters of the character string and the sum of the candidate ranks may be used.

The display method of the display unit may be changed by using one or more of different colors, different luminances, blinking, and underlining, for example, in addition to the above-described method.

The post-processing unit may perform the knowledge processing using the word information as described below in addition to the above. In order to check whether a word corresponding to a character string composed of candidate characters exists in the word information, the character string of the knowledge processing unit sentence is compared with the word of the word information, and the similarity between these character strings and words is determined. The degree or the degree of inconsistency is calculated. As a word corresponding to the character string, for example, a word whose similarity with the character string exceeds a predetermined threshold value or a word whose degree of mismatch with the character string does not exceed a predetermined threshold value is detected. (1) When a character string whose similarity exceeds a predetermined threshold value or a character string whose inconsistency degree does not exceed a predetermined threshold value is detected,
The maximum similarity or the minimum dissimilarity is detected from the detected character strings, and the word of the word information corresponding to the character string having the maximum similarity or the minimum dissimilarity is subjected to the knowledge processing result.
And the maximum similarity or the minimum dissimilarity is output as an evaluation value of the knowledge processing. (2) A character string whose similarity exceeds a predetermined threshold or a character string whose dissimilarity does not exceed a predetermined threshold even if all character strings for one unit of knowledge processing have been collated with words in word information.
If none of them can be detected, the result of the knowledge processing of the character string of the combination of the candidate characters having the first candidate rank and the predetermined lower limit of the similarity or the predetermined upper limit of the degree of inconsistency are evaluated. Output as a value. The lower limit of the degree of similarity and the upper limit of the degree of inconsistency indicate that the word corresponding to the character string composed of the candidate characters did not exist in the word information.

Further, in the above-described embodiment, an example of knowledge processing using word information has been described. However, the present invention may be applied to a character recognition device that performs knowledge processing using context information and other knowledge information.

[0027]

As described above, according to the present invention, when a correct answer is selected from candidate words, the display on the display unit is performed when the word of the word information originally existing in the knowledge dictionary is output as the candidate word. The case where the word of the word information registered in the word registration unit is output as a candidate word changes, and in the operation of selecting the candidate word, the word of the word information that originally existed in the knowledge dictionary and the user registered Since the word of the word information can be easily distinguished from the word information, the operator can instantly find a candidate word having a high possibility of a correct answer, and perform operations such as correction and confirmation. As a result, the time required for correcting or confirming the form or document is reduced. Therefore, it is possible to provide a character recognition device with good operability that can process a form or a document quickly and quickly.

[Brief description of the drawings]

FIG. 1 is a functional block diagram showing a configuration of an embodiment of the present invention.

FIG. 2 is a diagram illustrating an example of a form.

FIG. 3 is a diagram showing an example of candidate word selection.

[Explanation of symbols]

 Reference Signs List 10 character recognition device 12 recognition unit 14 word registration unit 16 post-processing unit 18 result editing unit 20 display unit 22 input unit 24 control unit 26 photoelectric conversion unit 28 image memory

Continuation of the front page (56) References JP-A-3-58182 (JP, A) JP-A-3-99386 (JP, A) JP-A-63-143685 (JP, A) JP-A-1-145789 (JP) , A) (58) Field surveyed (Int. Cl. ⁷ , DB name) G06K 9/72 G06K 9/03

Claims (1)

(57) [Claims] 1. A recognition unit for outputting a recognition result of a character pattern cut out from image data of a quantized form or document, a word registration unit for registering word information in a knowledge dictionary, and using the knowledge dictionary. A post-processing unit that outputs a knowledge processing result and a candidate word based on the recognition result, a result editing unit that edits the post-processing result, a display unit that displays the candidate word, and an input unit that inputs a correct word. In the character recognition device provided, the display unit may display the candidate word when the candidate word is a word of the word information originally existing in the knowledge dictionary, and when the candidate word is the word of the word information registered by the word registration unit. Wherein the display method of the word is changed.