Optical character recognition method and apparatus

• US 6,038,342 A
• Filed: 08/31/1993
• Issued: 03/14/2000
• Est. Priority Date: 08/10/1988
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Caere Corp. (Mexico) (Microsoft Corporation)
• Inventors: George, David, Higgins, Bryan, Dilworth, John, Bernzott, Phillip, Knight, Jeremy
• Primary Examiner(s): Couso, Jose L.
• Application Number:
  US08/114,608
• Time in Patent Office: 2,387 Days
• Field of Search: 382/14, 382/15, 382/25, 382/30, 382/34, 382/33
• US Class Current: 382/173
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/146 Aligning or centring of the image pick-up or image-field
• CPC Class Codes: G06V 30/414 Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text

Abstract

A system for recognition of characters on a medium. The system includes a scanner for scanning a medium such as a page of printed text and graphics and producing a bit-mapped representation of the page. The bit-mapped representation of the page is then stored in a memory means such as the memory of a computer system. A processor processes the bit-mapped image to produce an output comprising coded character representations of the text on the page. The present invention discloses parsing a page to allow for production of the output characters in a logical sequence, a combination of feature detection methods and template matching methods for recognition of characters and a number of methods for feature detection such as use of statistical data and polygon fitting.

Claims

1. A method for optically recognizing characters on a medium and storing a template of said recognized characters in a template cache, said template cache for recognition of subsequent characters on said medium, said method comprising the steps of:

(a) analyzing a first shape characteristic of a first character with a first shape characteristic analyzing process;

(b) analyzing a second shape characteristic of said first character with a second shape characteristic analyzing process;

(c) identifying said first character based upon the first and second shape characteristics;

(d) generating a template of said identified first character, and

(e) storing said template in said template cache.

2. The method as recited in claim 1 wherein said step of generating a template of said identified first character is further comprised of the steps of:

(a) generating a must be off pixel pattern, and

(b) generating a must be on pixel pattern.

3. The method as recited in claim 1 wherein said step of storing said template in said template cache is further comprised of the steps of:

(a) storing a pixel pattern of said identified character in said template cache;

(b) storing said must be off pixel pattern in said template cache;

(c) storing said must be on pixel pattern in said template cache, and

(d) storing a header in said template cache, said header for associating said pixel pattern, said must be off pixel pattern and said must be on pixel pattern.

4. In an optical character recognition system having a feature analysis process for identifying an unknown character, said optical character recognition system for identifying characters in a medium, a method for constructing a template library for use while processing said medium, said method comprising the steps of:

(a) identifying said unknown character with said feature analysis process;

(b) building a template for said unknown character subsequent to having identified said unknown character; and

(c) storing said template in said template library.

5. The method as recited by claim 4 wherein said template library is created for each new medium presented to said optical character recognition system.

6. The method as recited in claim 4 wherein prior to said step of identifying said unknown character with said feature analysis process, performing said step of determining that said unknown character does not match a template in said template library.

7. In a character recognition system having template matching as a first character recognition process, a method for generating a template cache for recognizing characters on a medium being processed, said method comprising the steps of:

a) providing a first pixel pattern of an unrecognized character on said medium being processed;

b) determining said unrecognized character on said medium being processed does not match a template in said template cache;

c) recognizing said unrecognized character utilizing a second character recognition process to create a recognized character;

d) building a template of said recognized character; and

e) adding said template of said recognized character to said template cache.

8. The method as recited in claim 7 wherein said step of building a template of said recognized character is further comprised of the steps of:

a) generating a second pixel pattern corresponding to pixels which must be off;

b) generating a third pixel pattern corresponding to pixels which must be on; and

c) generating a header for associating said first pixel pattern, said second pixel pattern and said third pixel pattern.

9. The method as recited in claim 8 wherein said second character recognition process is a feature analysis process.

10. A system for optically scanning a medium to recognize characters thereon, said system comprising:

scanning means for providing as output a bit-mapped image of said medium;

a memory means coupled with said scanning means, said memory means for storing said bit-mapped image;

a parsing means coupled to said memory means, said parsing means for extracting unknown characters from said bit-mapped image;

a template cache memory for storing templates of recognized characters;

a template character recognition means coupled to said template cache and said parsing means, said template character recognition means for recognizing characters by comparing unknown characters to templates in said template cache;

a feature analysis recognition means coupled to said template cache and said memory means said feature analysis recognition means for recognizing characters not recognized by said template character recognition means; and

template generation means coupled to said feature analysis recognition means and said template cache memory, said template generation means for generating a template of recognized character and storing in said template cache.

11. The system as recited in claim 10 wherein a template is comprised of:

a pixel pattern of a recognized character;

a must-be on pixel pattern of said recognized character; and

a must-be off pixel pattern of said recognized character.

12. The system as recited in claim 11 wherein said template generation means is further comprised of:

means for generating a must-be on bit pattern from a bit pattern of said recognized character; and

means for generating a must-be off bit pattern from said bit pattern of said recognized character.

13. A method of creating a character template using an image, said image including a representation of a character, said method including the steps of:

a) accessing said representation;

b) analyzing said representation using a first recognition process;

c) analyzing said scanned representation using a second recognition process, if said character is not recognized using said first recognition process;

d) generating a character template, if said character is recognized using said second recognition process, and

e) storing said character template, if a character template has been generated.

14. The method of claim 13 wherein said first recoginition process includes template matching and said second process includes feature analysis.

15. The method of claim 14 wherein said character template includes a first pixel pattern, a second pixel pattern, and a third pixel pattern, said first pixel pattern corresponds to said representation, said second pixel pattern includes a must-be-off pattern, and said third pixel pattern includes a must-be-on pattern.

16. A method of recognizing a first character in a character recognition system, said character recognition system including an image, said image including a first image representation corresponding to said first character, said method comprising the steps of:

a) accessing said first image representation;

b) applying a template matching process to said first image representation;

c) applying a feature analysis process to said first image representation, if said template matching process does not successfully identify said first image represenation as corresponding to said first character,

d) identifying said first image representation as corresponding to said first character, and

e) including a first character template corresponding to said first image representation for use in subsequent template matching processing.

17. The method of claim 16 wherein said image includes a second image representation corresponding to said first charater, and said method further comprises the steps of:

f) accessing said second image representation, and

g) applying said templating matching process, including said first character template, to said second image representation.

18. A system for recognizing a character represented on an image, said system comprising:

a) a memory for storing said image and a set of character templates, and

b) a processor being coupled to said memory, said processor for accessing said image and generating a first representation of a portion of said image, said portion including a representation of said character, said processor further for applying a templating matching process to said portion, said processor further for applying a feature analysis process to said portion if said template matching process does not identify said portion as corresponding to said character, said processor further for generating a character template from said portion if said feature analysis identifies said portion as corresponding to said character.