JPS61131091A - Character reader - Google Patents

Abstract

PURPOSE:To discriminate whether output data are post-processed or not and to clarify the discriminated result. CONSTITUTION:The recognition data in a 'sei' (a Chinese character pronounced by 'sei' and meaning 'famiry name') field are sent from a control part, a post-processing part 3 retrieves all 'sei' files, finds out the similarity between the sent recognition data and data retrieved from the 'sei' files and outputs the data having the maximum value of similarity and having a fixed value or more to a deciding part 4 as a correct answer. The deciding part 4 decides whether the data sent from the part 3 are post-processed and regarded as a correct answer or regarded as no-proposed character without post-processing and returns the data sent from the part 3 to a control part 1 together with the decided result. The control part 1 applied a high luminance display control code at the transfer to a CRT display device 5, a double printing control code at the transfer to a printer 6, or a control code for adding (*) to an 1-byte stand-by area formed on the head of the field at the transfer to a floppy disk device 7 to the data decided as no post-processed data by the deciding part 4.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a character reading device having a function of performing post-processing using knowledge engineering techniques when recognizing characters such as Chinese characters.

[Technical Background of the Invention] In recent years, with advances in H2J technology and element technology, character reading devices have become capable of reading Chinese characters.

However, compared to reading alphanumeric characters, reading these kanji requires extremely advanced technology due to the increased number of character types and similar characters. For this reason, in order to improve the overall reading accuracy of character reading devices that can read kanji, we not only improve the four so-called technologies for individual characters in optical character reading devices, but also improve the recognition results of individual characters (hereinafter referred to as recognition data). ) is used for post-processing using a so-called knowledge engineering method to check whether the words are appropriate as words or sentences.

[Problems with the Background Art] By the way, in a character reading device that uses post-processing as described above, the final output data includes data that has been post-processed with respect to individual character recognition data, and data that has not been post-processed. will be included.

However, in conventional reading devices, these output data have exactly the same output format, and no distinction is made between the two. In this way, if it is not possible to distinguish between post-processed data and non-post-processed data, the operator may have to check the data and find errors such as form entry errors, misreading of individual characters, deficiencies in post-processed data, etc. The problem was that it greatly reduced work efficiency when discovering and correcting problems.

[Object of the Invention] In view of the above-mentioned drawbacks, an object of the present invention is to provide a character reading device that can distinguish and clearly indicate whether or not output data has been subjected to post-processing.

[Summary of the Invention] The present invention includes a determining means for determining whether post-processing has been performed on the recognition results of individual characters, and road recognition output data is post-processing based on the results of the determining means. The above object is achieved by providing a control means for distinguishing and clearly indicating whether or not.

[Embodiment of the Invention] An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a character reading device of the present invention. An optical %T display 5, a printer 6, and a floppy disk device 7 are connected to the control section 1. Further, a database 8 is connected to the post-processing section 3 .

The control section 1 controls an optical Chinese character reading device 2, a post-processing section 3, a determining section 4, a CRT display 5, a printer 6, a floppy disk device 7, and the like. The optical Chinese character reading device 2 individually recognizes characters written on a form or the like one by one, and sends the results to the control section 1. The post-processing unit 3 calculates the degree of similarity between the individual recognition data output from the optical kanji reader 2 and the data stored in the database 8, and as a result, the data that has the maximum value and exceeds the threshold value Processing is performed to determine the correct answer. The determining unit 4 determines whether the individual recognition data output from the optical Chinese character reading device 1 has been processed by the subsequent processing unit 3. The database 8 is a storage device that stores words that may occur in each field of a form. It is assumed that the CRT display 5 is capable of high-X brightness display and the printer 6 is capable of double printing.

Next, the operation of this embodiment will be explained. First, the optical Chinese character reading device reads the characters on the form, performs individual recognition for each character, and sends the recognition data to the control section 1. Here, it is assumed that the characters shown in FIG. 2 are written on the read form. Therefore, among the received recognition results, the control unit 1 sends the recognition data for the first and last name and company name individually to the subsequent processing unit 3, and for the address, three fields (in this case, Tokyo, Ome City, Suehiro Town) will be sent with upper and lower relationships. In addition, the relationship between upper and lower ranks means, in a concrete example, that Tokyo is at the top, followed by Ome City, and Suehiro Town at the bottom.

The post-processing section 3 uses the data in the database 8 to process the sent individual recognition data in a knowledge-based manner. Here, the database 8 stores words that may occur in each field of a form having the contents shown in FIG. In other words, the "surname" file stores, for example, about 2,000 types of surnames with many occurrences and degrees, such as "Chinese" and "Sato," and the "given name" file stores, for example, about 4,000 types of surnames, such as "Obe" and "Hobe." In addition, the names of companies listed on the 1st and 2nd sections are stored in the "company name" file.

Further, the address file of the database 8 stores addresses all over Japan in a tree structure up to the third layer. however,
This tree structure up to the third layer means, for example, Tokyo is on the first layer, below which cities such as Ome City and Akishima City form the second layer, and below this second layer are Suehiro Town, Oyama Town, etc. This means that the towns are stored in a tree structure as the third layer.

When the recognition data of the "surname" field is sent from the control section 1 to the post-processing section 3, the post-processing section 3 searches all the "surname" files in the database 8 and combines the sent recognition data with "surname" field. The degree of similarity with the data retrieved from the "Last Name" file is determined, and as a result, the data that shows the maximum value and is greater than a certain value is output to the determination section 4 as the correct answer. If the maximum value does not exceed a certain value in the similarity processing of the post-processing unit 3, the post-processing unit 3 determines that there are no candidates and directly passes the individual recognition data sent from the control unit 1 to the determination unit 4.
send to The determining unit 4 determines whether the data sent from the downstream processing unit 3 has been subsequently processed and determined to be correct, or has not been subjected to downstream processing and has been determined to have no candidates, and sends the data to the downstream processing unit with this determination result. The data sent from 3 is sent back to the control unit 1. The control section 1 outputs the data sent back from the determination section 4 to the CRT display 5, printer 6, and floppy disk device 7. When the control section 1 outputs data to each output device, the control section 1 does not add any control code to the data determined by the determination section 4 to have undergone post-processing and sends the data to each output device as it is. However, the control unit 1 applies a high-intensity display control code when sending data to the CRT display device 5 and a double print control code when sending it to the printer 6 to the data determined by the judgment unit 4 as not to be processed later. , when sending to the floppy disk device 7, leave a 1-byte spare area at the beginning of the field and write "
A control code that adds "*" is added and sent.

FIG. 2 shows an example of a reading result displayed on the screen of the CRT display 5 of the character reading device shown in FIG. In this example, "Toshiba Ome Factory" in the company name field is displayed with high brightness, indicating that it has not been processed by the subsequent processing section 3.

According to this embodiment, it is possible to clearly indicate whether or not the output data has been subjected to post-processing.
In the case where the display device 5 displays a high-intensity display and the operator knows that a new town name Suehiro Town has been created, it is easy to discover that the database 8 does not contain data for Suehiro Town. This makes it easy to update the database. In addition, if Suehiro Town is displayed with high brightness as above, and the operator knows that this town name does not exist, it is easy to see that it is an incorrect entry on the form or a misreading by the optical character reader 2. It is possible to estimate the accuracy of the reading data, efficiently perform the verification work on the read data, and significantly improve the operability of the device.

Note that the present invention can be applied to a system equipped with a voice recognition device or a graphic recognition device instead of the optical Chinese character reading device shown in FIG. 1 to obtain the same effect.

[Effects of the Invention] As described above, the character reading device of the present invention includes a determining means for determining whether or not post-processing has been performed on the individual character recognition results in the optical kanji reading device, and output data that has been post-processed. By including the control means for explicitly outputting the output result, there is an effect that it is possible to distinguish and clearly indicate whether or not the output result has been subjected to post-processing.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the character reading device of the present invention, and FIG. 2 is a diagram showing an example of the display screen of the CRT display of FIG. 1. 1... Control unit 2... Optical Kanji reading device 3... Post-processing unit 4... Judgment unit 5...
CRT display 8...-Database agent Patent attorney Noriyuki Chika (and 1 other person) Figure 1 I.

Claims (1)

[Claims] In a character recognition device equipped with a function of post-processing the recognition results of individual characters, a determination means for determining whether post-processing has been performed on the recognition results of the individual characters, and output data based on the results of the determination means. 1. A character reading device comprising: a control means for distinguishing and clearly indicating whether or not a character has been subjected to post-processing.