JPH0636067A - Character reader - Google Patents

Abstract

PURPOSE:To provide the character reader which reduces the trouble of correcting a misrecognized character and shortens the correction period of time. CONSTITUTION:When the misrecognized character is designated with a cursor as shown in the figure while displaying on a screen, a group of character candidates for the misrecognized character is displayed in a window together with the similarities of the respective character candidates displayed in a bar graph. An operator while referring to the lengths of the bar graph can investigate character candidate with higher similarities mainly, and the correction of the character is facilitated to increase the selecting speed of the character.

Description

Detailed Description of the Invention

[0001]

The present invention relates to characters (including symbols).
A character reading device that obtains image data representing a character, recognizes each character included in an image based on the image data, and outputs a corresponding character code, and particularly corrects a character that is erroneously recognized. It is related to the screen display when doing.

[0002]

2. Description of the Related Art Character recognition in a well-known optical character reader (OCR) is performed as follows. First, a character image for each character is cut out from a document image obtained by optically reading a document with an image scanner, and the feature amount of the character is extracted from the cut out character image. The extracted feature amount is compared and collated with the stored contents of the recognition dictionary that stores the basic feature amount for each recognizable character, and the character code of the character having the most similar feature amount is used as the recognition result. Output. Further, a candidate character group composed of a plurality of candidate characters having similar feature amounts to some extent is also output.

By the way, since the recognition of characters is not always performed correctly, it is necessary to perform after-correction work for correcting an erroneously recognized character into a correct character. In addition, there are cases where it is desired to check candidate characters even if the characters seem to be correct. Such correction work is performed by displaying a recognition character as a recognition result on a display device and operating a command input device such as a keyboard or a mouse by a person.

FIG. 4 shows a conventional method for correcting misrecognized characters and the like. Among the recognized character strings displayed on the screen of the display device, by placing the cursor on a character that seems to be misrecognized and performing a click operation, a plurality of characters extracted in the recognition process of the recognized character are displayed in the window on the screen. Display the candidate character group. That is, in the example of FIG. 4, the character having the first degree of similarity is “gentle” which is a misrecognized character, and the character having the second highest degree of similarity is the second candidate character “God”. The third largest character is the third candidate character “Oshi”.

Since each of the candidate characters forming the displayed candidate character group is a character having a high degree of similarity,
It is highly possible that the correct character is included in this candidate character group. Therefore, the operator moves the cursor to the correct position of the candidate character "God" and performs a click operation at this position. Thereby, the erroneously recognized character can be corrected.

[0006]

By the way, when displaying a candidate character group as described above, since the candidate character and the degree of similarity or dissimilarity are not displayed, the operator cannot select which character. It is necessary to consider whether each is a correct character for each candidate character. Therefore, the correction takes time and burdens the operator. In addition, an operator's selection error is likely to occur.

Further, in the case of displaying the candidate character group as described above, when the candidate character includes a large number of characters having similar glyphs, it is difficult to make a difference from other characters only by displaying the candidate character. Will pay close attention to the selection of candidate characters. Therefore, if there is such a candidate character, the operator is burdened and selection mistakes are further increased. To give a concrete example, the katakana "ri"
And hiragana “ri”, katakana “he” and hiragana “he”, kanji “mouth” and katakana “lo”, kanji “ko” and katakana “e”, symbols "-(Minus)" and katakana "-(long sound)" and the symbol "-"
(Hyphen) "

Therefore, an object of the present invention is to solve the above-mentioned technical problems and to provide a character reading device capable of remarkably reducing the troublesomeness of the correction work of erroneously recognized characters and shortening the correction time. That is.

[0009]

A character reading apparatus according to claim 1 for achieving the above object recognizes a character represented by image data, and includes a plurality of characters including a recognized character which is a recognition result. Recognition means for outputting a candidate character group consisting of the candidate characters, a recognition character display means for displaying the recognition characters in the candidate character group output by the recognition processing means on a screen, and a recognition character display means for displaying the recognition characters. Character recognition means for indicating an arbitrary recognition character among the recognized characters with a cursor on the screen, and in response to an instruction input from this character designating means, the candidate character group corresponding to the character is displayed on the screen. A candidate character group display means for displaying, a selection input means for selecting a desired candidate character in the candidate character group with a cursor on the screen, and a newly recognized candidate character by replacing the selected candidate character with the character. Sentence And means for a candidate character group display means, the candidate character in the window of the screen, and displays with graphic corresponding to the similarity or dissimilarity of each candidate character.

According to this structure, when the desired character is designated by the character designating means while the recognized character is displayed on the screen, the candidate character group of this character shows the similarity or difference between the candidate characters represented by the graphic. It is displayed in the window on the screen along with the degree. As a result, the operator can grasp the similarity or difference of the candidate characters at a glance while looking at the figure, and this figure can be used as a judgment material when selecting the candidate character, and the candidate character can be quickly selected. It will be easier.

Further, in the character reading device according to the second aspect, the candidate character group display means includes the candidate characters in the window of the screen and the figures corresponding to the similarity or difference of each candidate character, and each candidate. The character attributes are also displayed. Such attributes include, for example, “character type”, “reading”, and “meaning when translated into another language” of the candidate character.
Can be given.

By thus displaying the candidate characters and the attributes corresponding to them in this way, the operator can easily select the candidate characters even when it is difficult to distinguish them from the character shapes alone. Further, the character reading device according to claim 3 is a means for displaying, for each recognized character displayed on the recognized character display screen, a graphic representing the similarity or difference of the character on the same screen in association with the character. Is further included.

According to this character reading device, for each recognized character displayed on the recognized character display screen, a graphic representing the degree of similarity or difference of the character is displayed, so that the operator reads and corrects the sentence. You can use this figure as a clue to find the character you need to correct, without having to search for the character you need. Therefore, the labor of the operator is reduced and the correction time is also shortened.

[0014]

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. FIG. 2 is a perspective view showing the overall configuration of an optical character reading device according to an embodiment of the present invention. The character reading device includes an image scanner 1, a Chinese character OCR unit 2 which is a main body for character recognition, and a Chinese character OCR unit 2.
While displaying the character string recognized by the screen,
It is composed of a personal computer or a workstation (hereinafter referred to as "personal computer") 3 for converting, editing and storing sentences.

FIG. 3 is a block diagram showing the basic construction of the character reading device. The image data obtained by the optical reading by the image scanner 1 is input to the recognition processing unit 21 of the Kanji OCR unit 2. Recognition processing unit 21
Includes a recognition dictionary that stores the basic feature amount of recognizable characters. In the recognition processing unit 21, a character image for one character is cut out from the input image data,
A feature amount is extracted from the cut out character image. Then, the extracted feature amount and the feature amount stored in the recognition dictionary are compared and collated, and the character having the highest degree of similarity or the lowest degree of difference is set as the recognized character that is the character of the recognition result. Further, a predetermined number (e.g., 8) of characters are output as a candidate character group in descending order of similarity or dissimilarity. A recognition character is also included in this candidate character group.

The candidate character group including the recognized character is stored in the candidate character buffer 31 of the personal computer 3.
Among the candidate characters stored in the candidate character buffer 31, the recognized character is the CRT through the display control unit 33.
6 is displayed on the screen. The CRT 6 is connected to the mouse 5 which functions as a character designating means and a selection inputting means. The mouse 5 is used for pointing a character or the like that is erroneously recognized among the displayed recognition characters (hereinafter referred to as “erroneously recognized character”).

When it is desired to convert a certain erroneously recognized character from the recognized character string appearing on the screen, the mouse 5 is operated to indicate the character. Then, on the screen of the CRT 6, a window display as shown in FIG. 1 appears just below the character "gentle" to be converted. And "gentle", "God",
A group of candidate characters including "push", ... Is displayed in the same column as the misrecognized character "gentle", and a bar graph showing the similarity of each candidate character is displayed at the same time. The length of the bar graph becomes longer as the similarity of the candidate character increases. Note that the candidate characters do not necessarily have to be all the candidate characters output from the recognition processing unit 21, and, for example, a character whose degree of similarity is lower than the reference value may not be displayed. The operator
When there is a character "God" to be converted in the candidate character group,
When the cursor is moved downward while looking at the bar graph and the mouse 5 is clicked in accordance with the character, the character is newly recognized as a recognized character and is replaced with a misrecognized character.

As described above, according to the present embodiment, when the erroneously recognized character is indicated by the cursor while the recognized character is displayed on the screen, the candidate character group of the erroneously recognized character is immediately adjacent to the erroneously recognized character. It is displayed together with the bar graph in the window that was set up. As a result, the operator can mainly consider candidate characters having a high degree of similarity, using the length of the bar graph as a criterion. Therefore, the characters can be corrected quickly, and the psychological and time burden on the operator is reduced.

In the above embodiments, the candidate character group is displayed together with the bar graph in the window, but other intuitive method can be used, for example, a certain number of figures (number depending on the degree of similarity). , Etc.) may be displayed together. Next, an embodiment corresponding to the invention described in claim 2 will be described. In the above-described embodiment, the operator visually searches for a character to be converted in the candidate character group with reference to the bar graph. However, as shown in FIG. 5, if you want to convert the character "ri" of "marine sports", if you operate the mouse 5 to indicate that character,
On the screen of the CRT 6, characters similar in shape such as “ri”, “ri”, “i”, and “″” appear immediately below the character “ri” to be converted, as shown in FIG. It is difficult to distinguish these characters while looking at a small screen, and since the glyphs are similar, there is not much difference in bar graph length.

Therefore, in this case, in order to distinguish between characters having similar glyphs, the attributes of each character are overlapped with the bar graph, that is, "Hiragana", "Katakana", "Symbol", "Kanji", "English character", "Numeral". Is displayed. As a result, the operator can easily select the candidate character even if it is difficult to distinguish the character shape alone, and it is possible to obtain the effects of reducing the operator's fatigue and reducing correction errors.

In the above-mentioned embodiment, the type of each character is displayed in Japanese while being superimposed on the bar graph, but it may be displayed in Roman characters or the like as shown in FIG. Further, reading of characters may be displayed as shown in FIG. The meaning of the characters may be displayed in a foreign language as shown in FIG. If you display it in Roman letters, etc., when you use the character reader outside Japan,
This is convenient when operated by an operator whose native language is not Japanese. Displaying the reading of characters has the advantage that similar Kanji can be quickly determined when they are lined up as candidate characters. Displaying the meaning of the characters in a foreign language is convenient when an operator who does not speak Japanese as a native language operates it, as described above.

Next, an embodiment corresponding to the invention described in claim 3 will be described. The bar graph display shown in FIG. 1 and FIG. 5 did not appear unless a character requiring correction was designated with a mouse or the like. Therefore, it is necessary to read the sentence and search for a character that needs to be corrected. In the present embodiment, the display control unit 33 determines, for each recognized character displayed on the screen, the degree of difference (for example, from 1) of the character. A figure representing a quantity that should be called the degree of anxiety when the character is selected) is displayed with a shadow on the background of the character (see FIG. 9).
FIG. 10 is an enlarged view showing such a background display. For example, “Gentle” of “Gentle Island”, “ground” of “ground”,
A large shadow (but not a black shadow, but a white frame is displayed for convenience of drawing) is displayed as a character with a high degree of anxiety in the "sphere" of the "music zone". "Island" in "Gentle Island", "Last" and "Raku" in "Last Music Area" have moderate anxiety.
"After" of "the last music circle" is a character with little anxiety,
A shadow is displayed for each. Characters without shadow display
It is a character that has been determined to have no anxiety.

Since the large, medium, and small shadows are displayed in the background in this way, the operator designates the character with the large shadow directly ahead of the character regardless of the connection before and after the sentence. Candidate character groups can be displayed in a window. Then, referring to the length of the bar graph and the attribute display of the characters, the mouse 5 is selected according to the character to be converted in the candidate character group.
When the click operation is performed, the character is newly recognized as a recognized character and is replaced with a misrecognized character.

In the above-described embodiment, the large, medium and small shadows are rectangular, but the shadows are not limited to this, and may be triangular as shown in FIG. 11, or as shown in FIG. Alternatively, it may be a vertically long stick. Although the embodiments of the present invention are as described above, the present invention is not limited to these embodiments. For example, in the above-described embodiment, the mouse 5 is used to indicate the erroneously recognized character and the candidate character, but the cursor key of the keyboard 7 or the like may be used.

In the above embodiment, the case where the image data is obtained by the optical reading has been described, but the image data may be obtained through a communication medium such as a facsimile machine. In addition, various design changes can be made without changing the gist of the present invention.

[0026]

According to the character reading device of the first aspect,
When an arbitrary character is designated by the character designating means while displaying the recognized character on the screen, the candidate character group of this character is displayed in the window together with the similarity or difference of each candidate character represented by the figure. .

As a result, the operator can mainly consider the candidate character having a higher degree of similarity or a lower degree of dissimilarity while referring to the figure, reducing the character correction load and speeding up editing. Can be faster. Claim 2
According to the described character reading device, the candidate character, the graphic corresponding to the similarity or difference of each candidate character, and the attribute of each candidate character are displayed together in the window of the screen. Even if the characters are similar to each other, the characters can be easily distinguished, and the operator can select the candidate character quickly.

According to the character reading device of the third aspect, the characters that need to be corrected are automatically displayed without considering the connection before and after the sentence. The burden is further reduced.

[Brief description of drawings]

FIG. 1 is a diagram illustrating a state in which a candidate character group is displayed together with a similarity of each candidate character represented by a bar graph in a window provided adjacent to a designated character.

FIG. 2 is a perspective view showing an entire character reading device.

FIG. 3 is a block diagram showing a basic configuration of a character reading device according to an embodiment of the present invention.

FIG. 4 is a diagram showing a display state of a conventional candidate character group.

FIG. 5 is a diagram showing another example in which the type of each character is displayed in an overlapping manner with a bar graph.

FIG. 6 is a diagram showing another example in which the type of each character is displayed in Roman letters or the like so as to be superimposed on a bar graph.

FIG. 7 is a diagram showing another example in which the reading of each character is displayed in an overlapping manner with a bar graph.

FIG. 8 is a diagram showing another example in which the meaning of each character is displayed in English overlaid on a bar graph.

FIG. 9 is a diagram showing a state in which a graphic representing the amount of anxiety when a recognized character is selected is displayed on the background of the character.

FIG. 10 is an enlarged view showing a background display of characters.

FIG. 11 is a diagram in which the shape of the background display of characters is changed.

FIG. 12 is a diagram in which the shape of the background display of characters is changed.

[Explanation of symbols]

 1 Image Scanner 2 Kanji OCR Unit 3 Personal Computer 5 Mouse 6 CRT 7 Keyboard 21 Recognition Processing Section 31 Candidate Character Buffer 32 Candidate Character Selection Section 33 Display Control Section

Claims (3)

[Claims] 1. A recognition processing means for recognizing a character represented by image data and outputting a candidate character group consisting of a plurality of candidate characters including a recognition character as a recognition result, and a candidate output by this recognition processing means. Recognized character display means for displaying the recognized character in the character group on the screen, and character instructing means for instructing any recognized character among the recognized characters displayed on the recognized character display means with a cursor on the screen. In response to an instruction input from the character indicating means, a candidate character group display means for displaying the candidate character group corresponding to the character on the screen, and a desired candidate character in the candidate character group on the screen. The selection character inputting means for selecting with the cursor and the means for replacing the selected candidate character with the character to be a newly recognized character, the candidate character group display means displays the candidate character in the window of the screen. , Character reader, characterized in that it is intended to display together with the figure corresponding to the similarity or dissimilarity of the candidate characters. 2. The candidate character group display means displays a candidate character, a graphic corresponding to the similarity or difference of the candidate character, and the attribute of the candidate character together in the window of the screen. The character reading device according to claim 1, which is characterized in that. 3. A means for displaying, for each recognized character displayed on the recognized character display screen, a graphic representing the degree of similarity or difference of the recognized character on the same screen in association with the character. The character reading device according to claim 1, which is characterized in that.