Apparatus for searching document images using a result of character recognition

• US 20020154817A1
• Filed: 09/12/2001
• Published: 10/24/2002
• Est. Priority Date: 04/18/2001
• Status: Active Grant

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Naoi, Satoshi, Katsuyama, Yutaka, Nishino, Fumihito
• Granted Patent: US 7,142,716 B2
• Time in Patent Office: Days
• US Class Current: 382/190
• CPC Class Codes: G06V 10/98 Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

A document image search apparatus generates a text by performing the character recognition of a document image and determines a re-process scope. Then, the apparatus generates a candidate character lattice from the re-recognition result of the re-process scope, generates character strings from the candidate character lattice and adds the character strings to the text. Then, the apparatus performs index search using the text with the character strings added.

Claims

1. A document image search apparatus, comprising:

an input device inputting a document image;

a character recognition device performing character recognition of the document image and generating a text of a recognition result;

a scope determination device determining a re-process scope of character recognition in the document image;

an addition device generating a text for search by adding a text obtained as a result of performing character recognition of the re-process scope again, to the text of the recognition result; and

a search device performing a search using the text for search.

2. The document image search apparatus according to claim 1, wherein

said scope determination device performs a language process for the text of the recognition result and extracts an area with low linguistic reliability as the re-process scope.

3. The document image search apparatus according to claim 1, wherein

said scope determination device extracts an area with low reliability of character recognition as the re-process scope using a reliability degree of the text of the recognition result.

4. The document image search apparatus according to claim 1, wherein

said scope determination device obtains an area where a reliability degree of the text of the recognition result is higher than a threshold value and where characters with high language process reliability consecutively continue in the text of the recognition result, and extracts an area other than the obtained area as the re-process scope.

5. The document image search apparatus according to claim 1, wherein

said scope determination device collates the text of the recognition result with a word dictionary and extracts an area with a specific morpheme as the re-process scope.

6. The document image search apparatus according to claim 5, wherein

said scope determination device extracts an area with at least one of a noun and an unregistered word as the area with the specific morpheme.

7. The document image search apparatus according to claim 1, wherein

said scope determination device collates the text of the recognition result with a word dictionary and extracts an area separated by a specific separator as the re-process scope.

8. The document image search apparatus according to claim 7, wherein

said scope determination device uses at least one of a boundary between character types and a punctuation mark for the specific separator.

9. The document image search apparatus according to claim 1, wherein

said scope determination device collates the text of the recognition result with a word dictionary, extracts an area with a specific morpheme and extracts a result obtained by extending the extracted area up to a specific separator as the re-process scope.

10. The document image search apparatus according to claim 1, wherein

said addition device includes:

a lattice generation device requesting said character recognition device to perform character recognition of the re-process scope and generating a candidate character lattice of the re-process scope; and

a character string generation device generating a character string from the candidate character lattice taking into consideration not only a first candidate character but also other candidate characters and outputting the character string as a text of a re-recognition result.

11. The document image search apparatus according to claim 10, wherein

said character string generation device collates the candidate character lattice with a word dictionary and generates the character string based on a word registered in the word dictionary.

12. The document image search apparatus according to claim 10, wherein

said character string generation device calculates distance values of paths obtained from the candidate character lattice, selects N higher-order paths in ascending order of the distance values of the paths and generates character strings that are determined by the selected paths.

13. The document image search apparatus according to claim 12, wherein

said character string generation device normalizes the distance values using a number of basic patterns, each of which is a minimum unit of character segmentation, and selects the N paths using the normalized distance values.

14. The document image search apparatus according to claim 10, wherein

said character string generation device calculates reliability degree of paths obtained from the candidate character lattice, selects N higher-order paths in descending order of the reliability degree of the paths and generates character strings that are determined by the selected paths.

15. A computer-readable storage medium which stores a program for enabling a computer to execute a process, the process comprising:

performing character recognition of a document image and generating a text of a recognition result;

determining a re-process scope of character recognition in the document image;

performing character recognition of the re-process scope again and generating a text of a re-recognition result; and

adding the text of the re-recognition result to the text of the recognition result and generating a text for search.

16. A propagation signal which propagates a program for enabling a computer to execute a process, the process comprising:

performing character recognition of a document image and generating a text of a recognition result;

determining a re-process scope of character recognition in the document image;

performing character recognition of the re-process scope again and generating a text of a re-recognition result; and

adding the text of the re-recognition result to the text of the recognition result and generating a text for search.

17. A document image search method, comprising:

performing character recognition of a document image and generating a text of a recognition result;

determining a re-process scope of character recognition in the document image;

performing character recognition of the re-process scope again and generating a text of a re-recognition result;

adding the text of the re-recognition result to the text of the recognition result and generating a text for search; and

performing a search using the text for search.

18. A document image search apparatus, comprising:

input means for inputting a document image;

character recognition means for performing character recognition of the document image and generating a text of a recognition result;

scope determination means for determining a re-process scope of character recognition in the document image;

addition means for generating a text for search by adding a text obtained as a result of performing character recognition of the re-process scope again, to the text of the recognition result; and

search means for performing a search using the text for search.