JPH10254871A - Document input method and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an efficient document inputting method and its device by clearly indicating an inputting position within an original and preventing the erroneous selection of a homonym. SOLUTION: Document original image data 501 is displayed on a picture 106a in parallel with the inputted document 502 of a document inputting person. Character recognition is executed with respect to characters in this document original and the recognizing result and a character string just after inputting by the document inputting person are collated with each other to search a coincident character string. A position in document image data 501 corresponding to this coincident character string is detected and the position 503 is displayed in data 501 in the picture to minimize the moving of the eyes of the document inputting person. For KANA (Japanese syllabary) and KANJI (Chinese character) conversion of the character string input of the document inputting person, at least one homonym candidates are outputted to collate the character string of the character recognizing result and the homonym candidates with each other and the homonym candidates are rearranged and outputted in a probable order based on the content of the coincident character candidate to prevent the erroneous selection of homonyms.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for realizing efficient document input by applying a character recognition technique for converting character or handwritten character image data into a character code.

[0002]

2. Description of the Related Art With the progress of OA, work of inputting a document as a character code string that can be processed by an information processing system has been performed in various fields. For this purpose, word processors (hereinafter word processors) have become common means.
For example, a method of inputting data from a keyboard while referring to a document by using such a method is widely used.

In this case, the input efficiency expressed by, for example, the amount of character code strings that can be input per unit time is most affected by the keying speed of the document input person, but is affected by the input environment such as the legibility of the original. Cannot be ignored. For example, in an input operation while referring to a densely described original, the line of sight of the document input person frequently reciprocates between the original surface and the word processing screen, and each time the input is performed within the original input. Since the position is lost, the input efficiency is reduced.

In addition, input of Japanese character string codes by a word processor or the like generally presupposes kana-kanji conversion. In this case, the information entered from the keyboard is
Although the reading of a character string described in a manuscript (kana), the presence of so-called homophone words in which different kanji notation character strings have the same reading, causes a reduction in input efficiency together with the above-mentioned loss of input position. I have.

[0005] That is, the document input person confirms whether or not the kanji-mixed character string (homophone) converted from the input "kana" matches the kanji-mixed character string described in the manuscript, and erroneously checks whether If it is converted, it must be corrected to the correct kanji mixed character string. For this reason, the number of times of line-of-sight reciprocation between the word processing screen and the document is increased, and the number of times of pressing the keyboard is also increased, so that the input efficiency is reduced.

On the other hand, as one technique for improving the efficiency of document input, a character recognition technique for converting character image data into a character code is expected. In character recognition technology,
The basis of the processing is to identify which character code the input character image data corresponds to based on the similarity with the character pattern that has been learned in advance. For this reason, the recognition performance is stable for printed characters having relatively little deformation of the font, and many products using this technology have been announced. However, if the character style unique to the writer is large for handwritten characters, especially handwritten Chinese characters, recognition errors frequently occur, and the effect of applying the character recognition technology is reduced.

[0007]

As described above, the conventional method of inputting a document using a word processor or the like has a problem that, apart from the ability of the document inputting person, the movement of the line of sight and the erroneous selection of the phonetic word reduce the input efficiency. Was. Further, the document input method using the character recognition technology has a problem that the effect of character recognition cannot be exerted when a character having large font deformation is recognized.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and has as its object to solve the above-mentioned problems in the prior art, to specify an input position in a manuscript, and to prevent a homonym selection error. And an efficient document input method and apparatus.

[0009]

In order to solve the above-mentioned problems, a character input method according to the present invention is a document input method using a character recognition technology, wherein image data of a document to be input is displayed on a screen. Character recognition is performed on characters in the input document, and a character code string input by the document input person is collated with a character code string that is the result of the character recognition to search for a matching character code string. Performing the search, detecting the position in the image data of the input target document corresponding to the matched character code string,
The detected position is displayed in the screen.

The character code string input by the document input person is subjected to Kana-Kanji conversion to obtain one or more homophone candidates, and the character code string which is the result of the character recognition and the homophone word are output. Matching with candidates is performed, and based on the content ratio of matching character candidates, the obtained homonym candidates are sorted in order from the most probable candidates.

According to another aspect of the present invention, there is provided a character input device using a character recognition technique, comprising: means for displaying image data of an input target document on a screen; Means for performing character recognition for the characters in the character string; means for comparing a character code string input by a document input person with a character code string that is the result of the character recognition; Means for searching for a code sequence, and a result of the search;
Means for detecting a position in the image data of the input target document corresponding to the matched character code string, and means for displaying the detected position on the screen.

In the above character input device, the character code string input by the document input person may be
Means for performing Kana-Kanji conversion to obtain one or more homophone candidates; means for collating a character code string as an execution result of the character recognition with the obtained homophone candidates; Means for rearranging the obtained homophone candidates in order from the most probable candidates based on the content ratio of the character candidates to be performed.

[0013] In the document input method and apparatus according to the present invention, image data of a document to be input is displayed on a screen, character recognition is performed on characters in the document to be input, and characters input by a document input person are performed. Code strings,
By comparing the character code string which is the result of the character recognition with the character code string, searching for a matching character code string, detecting a position in the document image data corresponding to the matching character code string, and setting the detected position on the screen. , The document input person can determine the input position with minimum movement of the line of sight. In addition, regarding the input of a character code string by the document input person, one or more homophone candidates are obtained by performing kana-kanji conversion, and the character code string that is the result of character recognition is collated with the homophone candidate to match. By rearranging the homophone candidates in order from the most probable candidates based on the content ratio of the homophone candidates that contain more character candidates, even if the current technology level of character recognition technology is used, it is possible to reduce the homophone selection error that occurs during kana-kanji conversion. Be prevented. These make it possible to perform efficient document input according to the ability of the document input person.

[0014]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a processing block diagram of a document input method and apparatus showing a first embodiment of the present invention. FIG.
101, document original image data to be input; 102, character recognition means for converting character image data into character codes; 103, a document inputter; 104, a means for inputting a Japanese character string; Input position detecting means 106 for detecting an input position, means 106 for displaying a document original image and a character code string input by the document input person 103, and 107 an input document file.

The detailed operation of the processing block shown in FIG. 1 will be described below.

First, the character recognizing means 102 performs a character recognizing process on the document original image data 101. Various methods have been proposed for character recognition processing.
Either method can be applied as the character recognition means used in the present invention. For example, as a specific method of the character recognition means, it can be realized by executing the processing steps of extracting a character area, extracting single character image data, and single character recognition. The character recognition unit 102 outputs a character code as a character recognition result and a character recognition result 108 including position information of each single character image data in the document original image data on which the character recognition has been performed.

Next, the document input user 103 inputs a document while referring to the document original image displayed by the document image / character string display means 106. This document input is performed using the Japanese character string input unit 104.
As the Japanese character string input means 104, various Japanese input means having a kana-kanji conversion function can be used. That is, the Japanese character string input unit 104 converts a kana character string input by a document input person into a character string mixed with kanji.
Various methods such as the longest match method and the minimum number of clauses method can be used as an algorithm for converting a kana character string into a character string mixed with kanji. A kana-kanji conversion library such as MS-IME can be used as a product level software package. It is also possible to use it. Japanese character string input means 104
Is a kanji-mixed character string 109 that is the result of the kana-kanji conversion. The unit of Kana-Kanji conversion depends on the operation contents of the document input person, and can be converted in various units such as single Kanji, word, phrase, phrase, sentence, and the like.

Next, in the input position detecting means 105, a character recognition result 108 output from the character recognizing means 102 is output.
Is compared with a character string 109 mixed with kanji output from the Japanese character string input means 104, and a character recognition result matching the character string mixed with kanji is searched for, thereby obtaining the character currently input by the document input unit 103. Detect the position. The input position detecting means 105 outputs the detected input position information included in the information 110. In the document image / character string display means 106, the document original image data 101,
A character string 109 mixed with kanji input by the document input person 103 and information 110 including a character position in a currently input document image are displayed on the screen. here,
Various methods are conceivable for displaying the character position currently being input. For example, a method of reversely displaying the end of the document original image data corresponding to the last input kanji mixed character string can be used.

In the document input according to the present invention described with reference to FIG. 1, the document input is performed by adjusting the line of sight to the input position displayed on the document original image screen.
Unnecessary gaze movement can be prevented. FIG.
The details of the processing of the input position detecting means 105 and the character image / character string displaying means 106 in the description will be described later with reference to FIGS.

FIG. 2 is a diagram showing a second embodiment of the present invention. In the second embodiment of the present invention shown in FIG. FIG. 10 is a processing block diagram of a document input method and a device when “.” Is added. In FIG. 2, reference numeral 201 denotes the added homonym rearranging means. The processing contents of the document input method and the apparatus according to the embodiment shown in FIG. 2 are different from those of the document input method and the apparatus according to the first embodiment shown in FIG. 1 except for the differences described below. , All have the same processing content.

The first difference is that the Japanese character string input means 104 outputs a plurality of different kanji mixed character strings for the same reading (kana). For example, for the reading of "Ito", the method of FIG. 1 outputs only the kanji character string of Ito, while the kana-kanji conversion input means 104 shown in FIG. 2 uses a plurality of characters such as Ito, Ito, Ito, etc. Is output.

The second difference is that the homonym rearranging means 201 uses the homonym candidate 109 output from the Japanese character string input means 104 and the character recognition result (output) from the character recognition means 102. (Character candidate) 202, and
The point is that the homophone candidates are sorted in order from the most probable candidates based on the content ratio of the character candidates matching the character candidates. Therefore, the output from the homonym rearranging means 201 can be a homonym candidate with high reliability.

As described above, by adding the homonym rearranging means 201, it is possible to output more likely homonym candidates, and at the same time, it is possible to improve the accuracy of input position detection.

FIG. 3 shows the input position detecting means 105 in FIG.
FIG. 3 is a diagram for describing in detail the configuration and processing contents of FIG.

In FIG. 3, reference numeral 301 denotes the character recognizing means 10.
Character recognition result in information 108 output from 2
2 is the information 10 similarly output from the character recognition means 102
8 is character recognition position information. Reference numeral 303 denotes an input character string which is the output of the Japanese character string input unit 104 shown in FIG. 1 or the output of the homophone rearranging unit 201 shown in FIG. The processing unit of this input character string is shown in FIG.
The unit is equivalent to the unit input by the Japanese character string input unit 104 in FIG. 2 or FIG. Reference numeral 304 denotes position information of an input character string which is an output of the Japanese character string input unit 104 shown in FIG. 1 or an output of the homophone rearranging unit 201 shown in FIG. The position information of the input character string is also position information on the character string of the unit input by the Japanese character string input means 104 in FIG. 1 or FIG. Information. A character candidate 305 is a character recognition result.
This is a character position matching unit that matches the input character string 303 with the input character string 303 and searches for a matching character string. When a combination of a matching character candidate and an input character string is found by the character position matching, the character position matching unit 305 determines the corresponding character image data position information of the character candidate in the document manuscript image data, It outputs information 110 that is combined with the position information of the input character string in the character string set.

FIG. 4 is a diagram for explaining in detail the configuration and processing contents of the character image / character string display means 106 shown in FIGS. 1 and 2.

In FIG. 4, reference numeral 401 denotes document original image data 101 and current input character string position information 110.
This is a means for superimposing the position designation image data based on. Reference numeral 402 denotes a unit for actually displaying document manuscript image data and input character strings, and a general display device can be used.

The processing contents of the document image / character string display means 106 will be described below with reference to FIG. First, the superimposing means 401 matches the document manuscript image data with the position information A in the input character string set of the character string immediately after the input, which is the output from the character position collating means 305, and matches the character string immediately after the input. The position information B in the document original image data corresponding to the character recognition result is input. next,
An image for specifying the character string immediately after the input indicated by the position information A is synthesized. For this image synthesis, for example, various methods such as inversion of the target character string, underlining, enclosing, and the like can be applied. Next, an image for specifying an area in the document original image data indicated by the position information B is synthesized. Similarly to the above, a method of inverting, underlining, enclosing, and the like the target area image data can be applied to this image synthesis.

FIG. 5 is a diagram for explaining a display example of a document original image and an input character string on the display device screen in the above embodiment.

In FIG. 5, reference numeral 106a denotes a screen of the display device; 501, a document manuscript image; 502, an input document; 503, an input target area in the document manuscript image 501; Is shown.

In this display example, the original document is displayed as it is on the left side of the screen 106a as the image 501. on the other hand,
The document input and entered by the document inputter is displayed on the right side of the screen 106a, and the input character string is compared with the character recognition result of the document original image data by internal processing. Is set in the input target area 503 in the document original image 501.
Is displayed in a form that is easy for the document input person to see, and the document input person can determine the input position with minimum movement of the line of sight. Although the display example of FIG. 5 shows an example in which the document original image and the input document of the document input person are displayed side by side, they may be displayed vertically.

The first candidate of the character string converted by the kana-kanji conversion is displayed at the character string position 502 immediately after the input, and
Each time the conversion key is pressed, the candidates below the candidate are sequentially replaced with the preceding converted character string. Alternatively, the second and subsequent candidates may be displayed in a predetermined window frame, and may be replaced with the previous converted character string by moving the cursor using a conversion key or by designating a number. In short, by improving the likelihood of the first candidate,
In addition to improving the input efficiency by reducing the operation of selecting homophones, the use of character recognition technology at the current technical level prevents homonym selection errors occurring during kana-kanji conversion.

[0034]

As described above in detail, according to the present invention, the document input person can determine the input position with minimum movement of the line of sight. Even with the use, it is possible to prevent homonym selection errors that occur during kana-kanji conversion, which has the remarkable effect of enabling efficient document input according to the capabilities of the document input person. is there.

[Brief description of the drawings]

FIG. 1 is a processing block diagram for explaining the configuration and processing contents of a document input method and apparatus according to a first embodiment of the present invention.

FIG. 2 is a processing block diagram for explaining the configuration and processing contents of a document input method and apparatus according to a second embodiment of the present invention, wherein a homophone rearranging means is provided in the configuration shown in FIG. It is an added processing block diagram.

FIG. 3 is a processing block diagram for describing in detail a configuration and processing contents of an input position detection unit in the embodiment.

FIG. 4 is a processing block diagram for explaining in detail the configuration and processing contents of a document image / character string display unit in the embodiment.

FIG. 5 is a diagram for describing a display example of a document original image and an input character string on a display device screen in the embodiment.

[Explanation of symbols]

 101: Document manuscript image data 102: Character recognition means 103: Document input person 104: Japanese character string input means 105: Input position detection means 106: Document image / character string display means 201: Homophone rearrangement means 202, 301 ... Character recognition result 302 Character recognition position information 303 Input character string 304 Position information of input character string 305 Character position collating means 401 Display data superimposing means 402 Display means

 ──────────────────────────────────────────────────の Continued on the front page (72) Inventor Masami Oguro 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Inside Nippon Telegraph and Telephone Corporation

Claims (4)

[Claims] 1. A document input method using a character recognition technology, comprising displaying image data of an input target document on a screen, performing character recognition on characters in the input target document, and inputting the input data by a document input person. A character code string is compared with a character code string that is the result of performing the character recognition, and a search is made for a matching character code string. As a result of the search, the input target document corresponding to the matched character code string Detecting a position in the image data, and displaying the detected position on the screen. 2. A character code string input by the document input person is subjected to Kana-Kanji conversion to obtain one or more homophone candidates, and a character code string which is a result of the character recognition and the homophone word are obtained. 2. The document input method according to claim 1, wherein matching is performed with candidates, and the obtained homophone candidates are rearranged in ascending order based on the content ratio of matching character candidates. 3. A document input device using a character recognition technology, comprising: means for displaying image data of an input target document on a screen; means for performing character recognition on characters in the input target document; Means for collating the character code string input by the above with the character code string which is the result of the character recognition; means for searching for a matching character code string as a result of the collation; A document input device comprising: means for detecting a position in the image data of the input target document corresponding to the determined character code string; and means for displaying the detected position on the screen. 4. A means for performing Kana-Kanji conversion on a character code string input by the document input user to obtain one or more homophone candidates, and a character code string which is an execution result of the character recognition. Means for collating with the same homonym candidate, and means for rearranging the obtained homonym candidates in order from the most probable candidates based on the content ratio of the matching character candidates as a result of the collation. The document input device according to claim 3, wherein