Character recognition device and a method therefore

• US 20020154815A1
• Filed: 03/29/2002
• Published: 10/24/2002
• Est. Priority Date: 03/30/2001
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Digital Solutions Corporation (Toshiba Corporation)
• Original Assignee: Toshiba America Information Systems Incorporated (Toshiba Corporation)
• Inventors: Mizutani, Hiroyuki
• Granted Patent: US 7,233,697 B2
• Time in Patent Office: Days
• US Class Current: 382/181
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/148 Segmentation of character regions

Abstract

The present invention relates to an optical character recognition device (OCR) for reading a form provided with character frames in reading fields, into which a user fills each character. Characteristic vectors are extracted from the character images of each frame. A number of characters decision unit 16, into which the characteristic vectors are input, decides the number of characters filled in one of the character frames. A character separation unit 18 separates each of characters from the character image based on the number of characters decided by the decision unit 16. The character recognition unit 20 then recognizes each of the character. The OCR according o the present invention is able to read the form correctly, in which a plurality of characters are filled in one of the frames.

Claims

1. A character recognition device for recognizing characters filled in a form provided with character frames, comprising:

means for deciding a number of characters filled in each one of said character frame from a character image of said character frame, and

means for separating and recognizing each of the characters based on the number of characters decided by said decision means.

2. A character recognition device according to claim 1, wherein said means for deciding a number of characters further comprises means for obtaining a characteristic vector of said character image of said character frame and means for deciding said number of characters using a discriminant function of the number of characters, which is provided with said characteristic vector as an input, wherein said means for recognizing a character recognizes characters using said characteristic vector.

3. A character recognition device according to claim 2, wherein said discriminant function of the number of characters is a n-character discriminant function having a first function value when said number of characters filled in said each one of the character frames is n (n is an optional integer) and having a second function value when said number of characters filled in said each one of the character frames is not n.

4. A character recognition device according to claim 3, wherein said discriminant function of the number of characters is a one-character discriminant function having said first function value when said number of characters filled in said each one of the character frames is 1 and having said second function value when said number of characters filled in said each one of the character frames is plural.

5. A character recognition device according to claim 4, wherein said discriminant function of the number of characters further comprises a two-character discriminant function having said first function value when said number of characters filled in said each one of the character frames is 2 and having said second function value when said number of characters filled in said each one of the character frames is 3 or more.

6. A character recognition device according to claim 4 or 5, wherein said characters are numerals of 0 to 9.

7. A character recognition device according to claim 1, wherein said deciding means further comprises means for obtaining a first distribution of characteristic vectors of said character image, in which said number of characters filled in said each one of the character frames is n (n is an optional integer), means for obtaining a second distribution of characteristic vectors of said character image, in which said number of characters filled in said each one of the character frames is not n, means for inputting said characteristic vectors of said character image filled in said each one of the character frames, and means for deciding whether said number of characters of said characteristic vectors of said input character image is n or not based on the a distance between said input characteristic vectors and said first and second distribution of said characteristic vectors.

8. A character recognition device according to claim 1, wherein said means for deciding further comprises means for obtaining a first distribution of characteristic vectors of a plurality of character images composed of one-character filled in said each one of the character frames, means for obtaining a second distribution of characteristic vectors of a plurality of character images composed of 2 or more characters filled in said each one of the character frames, means for inputting characteristic vectors of the character images, with which said number of characters filled in said each one of the character frames is unknown, and means for deciding whether said number of characters of said characteristic vectors of said input character image is 1 or not based on the a distance between said input characteristic vectors and said first and second distribution of said characteristic vectors.

9. A character recognition device according to claim 1, wherein said means for deciding further comprises means for obtaining a first distribution of characteristic vectors of a plurality of character images composed of two-characters filled in said each one of the character frames, means for obtaining a second distribution of characteristic vectors of a plurality of character images composed of 3 or more characters filled in said each one of the character frames, means for inputting characteristic vectors of character images, with which said number of characters filled in said one-character frame is unknown, and means for deciding whether said number of characters of said characteristic vectors of said input character image is 2 or not based on the a distance between said input characteristic vectors and said first and second distribution of said characteristic vectors.

10. A character recognition device according to claim 8 or 9, wherein said characters are numerals of 0 to 9.

11. A character recognition method for recognizing characters filled in a form provided with character frames, comprising steps of:

inputting image information of a form provided with character frames, in which said characters are filled,

obtaining characteristic vectors of a character image in said each one of the character frames using said input image information,

deciding a number of characters in said each one of the character frames based on said characteristic vectors obtained,

separating characters one by one from said character image in said each one of the character frames based on the number of characters decided by said step, and

recognizing said characters separated by said step using said characteristic vectors.

12. A character recognition method according to claim 11, wherein said step of deciding said number of characters has a one-character decision step of deciding whether a number of characters filled in said each one of the character frames is 1 or more than 1.

13. A character recognition method according to claim 12, wherein said step of deciding said number of characters further comprises a two-character decision step of deciding whether said number of characters filled in said each one of the character frames is two or more than two when said number of characters filled in said each one of the character frames is decided as more than 1 by said one-character decision step.

14. A character recognition method according to claim 13, wherein said step of deciding said number of characters further comprises a three-character decision step of deciding whether said number of characters filled in said each one of the character frames is three or more than three when said number of characters filled in said each one of the character frames is decided as more than three by said two-character decision step, and wherein said step of deciding said number of characters increases sequentially said number of characters up to a predetermined upper limit number of characters, thereby repeating the steps of deciding said number of characters.

15. A character recognition method according to claim 12, wherein said one-character decision step further comprises a step of obtaining a first distribution of characteristic vectors of a plurality of character images composed of one-character filled in said each one of the character frames, a step of obtaining a second distribution of characteristic vectors of a plurality of character images composed of two or more than two-characters filled in said each one of the character frames, a step of inputting said characteristic vectors of said character image, a number of characters of which is not known, filled in said each one of the character frames, and a step of deciding whether said number of characters of said input characteristic vectors is one or not based on a distance between said input characteristic vectors and said first and second distribution of said characteristic vectors.

16. A character recognition method according to claim 12, wherein said one-character decision step further comprises a step of obtaining a first distribution of characteristic vectors of a plurality of character images composed of one-character filled in said each one of the character frames, a step of obtaining a second distribution of characteristic vectors of a plurality of character images composed of three or more than three characters filled in said each one of the character frames, a step of inputting said characteristic vectors of said character image, a number of characters of which is not known, filled in said each one of the character frames, and a step of deciding whether said number of characters of said input characteristic vectors is two or not based on the a distance between said input characteristic vectors and said first and second distribution of said characteristic vectors.

18. A computer program for executing said method stated in claim 12 or 13.

17. A computer program for recognizing characters filled in a form provided with character frames, wherein said program comprises functions of;

deciding a number of characters filled in one-character frame, and

separating and recognizing characters of said number of characters from a character image of said character frame based on the number of characters decided by said deciding function.