JPS61198383A - Character correcting device - Google Patents

Abstract

PURPOSE:To improve remarkably the read processing efficiency by executing correction processing when a character of recognition disable takes place in the recognizing processing. CONSTITUTION:The operator observes the result of recognition displayed on a picture of a display section 24 and it is discriminated that a character 25 differs from a substantial character 26 being read object and the mis-recognition is discriminated, a Kana (Japanese syllabary) for a correcting character to the mis-read character 25 is inputted by operating an input key of an operating section 23 and further, a Kanji (Chinese character converting key is operated. Then a control section 20 reads the corresponding correcting character displayed on the display section 24, the operator operates the operating section 23 to input confirming information and registration indication information for correcting character. The control section 20 gives a label to the inputted correcting character and registers it to a correction character registration buffer 22 in the state of a Kanji code. The control section 20 outputs the character of the inputted correcting character to, e.g., a computer as the final recognition result.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a character correction device used particularly for Kanji OCR.

[Technical Background of the Invention and Problems thereof] In recent years, optical character reading devices (OCR) have become highly functional devices whose readable double characters have been expanded from alphanumeric characters and kana characters to kanji (Kanji 0CR). ) is being realized. Kanji is a huge number of characters, 6349 characters including JIS level 1 and level 2, but with normal kanji OCR, the number of characters to be read is 2000 to 3000 characters due to capacity limitations of recognition dictionary memory and character appearance frequency. .

For this reason, in conventional Kanji OCR, a Kanji character other than the original target to be read, that is, a character that does not exist in the recognition dictionary memory, may be input. In this case, the kanji OCR results in a miss or misreading. Furthermore, even in the case of Chinese characters that exist in the recognition dictionary memory, characters that cannot be recognized may appear in OCR, which reads handwritten characters, due to the scribe's habits.

Conventionally, when an unrecognizable character occurs, an operator inputs a corrected character using a character correction device. Specifically, it is assumed that characters on a sheet of paper scanned by the reading unit 10 as shown in FIG. 3 are recognized by the control unit [11], and then the recognition result is displayed on the display unit 12. The operator checks the recognition result displayed on the screen of the display section 12, and in the case of misreading or misreading, operates the operation section 13 to perform correction processing.

In this case, the operator inputs kana character information and conversion instruction information for instructing kanji conversion from the operation unit 13 to the control unit @11. The control device 11 reads corrected characters from a character correction dictionary memory prepared in advance according to input information, and displays them on the display unit 12. Thereafter, when confirmation information is input from the operator, the control device 11 outputs the corrected characters (kanji code) as the final recognition result.

However, in the conventional character correction processing as described above,
If the number of occurrences of unrecognizable characters increases, the time required for correction operations by the operator increases.

Therefore, there is a problem that the OCR reading processing efficiency is reduced.

[Object of the Invention] An object of the present invention is to provide a character that, when an unrecognizable character occurs during recognition processing, can be reliably corrected by a simple operation, thereby greatly improving the reading processing efficiency. The purpose of this invention is to provide a correction device.

[Summary of the Invention] The present invention includes input means for inputting a corrected character and registration instruction information for instructing registration of the corrected character when an unrecognizable character occurs. The registration control means registers the corrected character in the corrected character memory in accordance with the registration instruction information inputted from the input means.

Next, when an unrecognized character occurs, if search instruction information for searching for a registered corrected character is input, the search control means searches the corrected character memory according to the search instruction information and displays the corrected character. Display it on the means. The modification control means executes modification processing based on the modification character input from the input means or the modification character specified among the characters searched by the search control means.

With such a character correction device, unrecognizable characters can be corrected by a simple input operation by an operator.

[Embodiment of the Invention] An embodiment of the present invention will be described below with reference to the drawings. 1st
The figure is a block diagram showing the configuration of a character correction device according to an embodiment. In FIG. 1, a control section 20 includes an OCR recognition section and controls the entire character correction apparatus. The character correction dictionary memory 21 stores in advance character codes including kanji characters required for correction. The modified character registration buffer 22 is a memory in which specified modified characters (character codes) are registered under the control of the control unit 20. Operation unit 2
Reference numeral 3 denotes a keyboard provided with input keys for inputting corrected characters, information for instructing registration of corrected characters, information for instructing search for registered corrected characters, and the like. The display unit 24 is, for example, a CRT display device that displays recognition results, corrected characters, etc. under the control of the control unit 20.

In the character correction device having such a configuration, the operation of the same embodiment will be explained. First, it is assumed that a sheet of paper to be read is scanned by a reading section (see FIG. 3), and characters recorded on the sheet are converted into image data and then transferred to the control section 20.

The control unit 20 executes Wgl processing on the characters on the paper as shown in FIG. 2(a), and outputs a recognition result as shown in FIG. Here, recognition processing is normally executed for each character.

The operator looks at the recognition result displayed on the screen of the display unit 24 and determines that the character 25 shown in FIG. 2(b) is different from the original character 26 to be read and has been misread. Based on this determination, the operator starts character correction processing. That is,
The operator inputs the corrected kana character for the misread character 25 by operating the input key on the operation unit 23, and further operates the kanji conversion key. As a result, the control unit 20
reads the corresponding correction character from the character correction dictionary memory 21 and causes the display unit 24 to display the corresponding kanji.

After confirming the corrected characters displayed on the display unit 24, the operator operates the operating unit 23 to input confirmation information and information for instructing registration of the corrected characters. The control unit 20 labels the corrected characters input as described above according to the registration instruction information, and then stores them in the corrected character registration buffer 2 in the state of kanji code.
Register for 2. The control unit 20 outputs the character code of the input corrected character to, for example, a computer or the like as a final recognition result.

Next, suppose that when a new character is scanned and recognition processing is performed, the same character as above is misread.

That is, assume that the control unit 20 displays the misread characters 25 shown in FIG. 2(b) on the display unit 24. After checking the screen of the display unit 24 for misreading, the operator operates the operation unit 23 to input search instruction information for registered correction characters. In response to this search instruction information, the control section 20 searches the corrected character registration buffer 22, reads out the registered corrected character candidates, and displays them on the display section 24. At this time, the control unit 20 displays all of the corrected characters registered in the corrected character registration buffer 22, but the number of displayed characters is small because characters that are likely to be misread or rejected are limited.

The operator selects the corresponding corrected character from among the corrected character candidates displayed on the display unit 24. Thereby, the control unit 20 outputs the selected corrected character as the final recognition result. Here, if there is no corresponding corrected character among the registered corrected characters displayed on the display section 24 in accordance with search instruction information input by the operator, the above-mentioned corrected character is input. That is, the operator inputs the corrected character and its registration instruction information to the control section 20 from the operation section 23. This modified character will be newly registered in the modified character registration buffer 22.

In this way, when an unrecognizable character that is misread or rejected by the recognition process appears, the corresponding corrected character is input and the corrected character is registered in the memory.

Next, when an unrecognizable character appears, the corresponding corrected character is retrieved from the registration memory and output. in this case,
The operator simply operates the input key for inputting search instruction information and selects the corresponding character from among the displayed correction characters. Therefore, it is possible to correct unrecognized characters by a simple input operation. If the corresponding corrected character is not registered in the registration memory, it is necessary to input the corrected character and registration instruction information, but normally unrecognizable characters are often limited to specific characters. Therefore, by registering a predetermined number of corrected characters in the registration memory, character correction processing can be performed with only the simple input operation as described above.

[Effects of the Invention] As described in detail above, according to the present invention, I! When an invalid character occurs,
Reliable correction processing can be performed with simple input operations. Therefore, even when unrecognized characters are relatively likely to occur when the characters to be read are kanji etc., the input operation for correction processing by the operator can be simplified, resulting in a significant improvement in reading processing efficiency compared to conventional methods. It is something that can be done.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a character correction device according to an embodiment of the present invention, FIG. 2(a) is an example of a character to be read, and FIG. 2(b) is a diagram showing an example of a recognition result. FIG. 3 is a block diagram showing the basic configuration of a conventional OCR. 20... Control unit, 21... Dictionary memory for character correction,
22...External pressure character registration buffer, 23...Operation unit,
24...Display section.

Claims (1)

[Claims] In an optical character reading device that recognizes characters recorded on paper, when an unrecognized character occurs, a first step is to input a corrected character corresponding to the corresponding character and registration instruction information for instructing registration of the corrected character. an input means, a registration control means for registering the corrected character in a corrected character memory in accordance with registration instruction information inputted from the first input means, and a registration control means for registering the corrected character to be registered when an unrecognized character occurs. a second input means for inputting search instruction information for a search instruction; and a means for displaying the corrected characters read out by searching the corrected character memory in accordance with the search instruction information input from the second input means. a search control means for performing a correction process based on a correction character input from the first input means or a correction character specified among the characters searched by the search control means; A character correction device comprising: