Candidate Identification by Image Fingerprinting and Model Matching

• US 20130051607A1
• Filed: 08/29/2011
• Published: 02/28/2013
• Est. Priority Date: 08/29/2011
• Status: Active Grant

• Current Assignee: DST Technologies, Inc. (SS&C Technologies Holdings Incorporated)
• Original Assignee: DST Technologies, Inc. (SS&C Technologies Holdings Incorporated)
• Inventors: Highley, Joshua O.
• Granted Patent: US 8,831,350 B2
• Time in Patent Office: Days
• US Class Current: 382/100
• CPC Class Codes: G06V 10/242 by image rotation, e.g. by 90 degrees
• CPC Class Codes: G06V 10/754 involving a deformation of the sample pattern or of the reference pattern; Elastic matching
• CPC Class Codes: G06V 10/768 using context analysis, e.g. recognition aided by known co-occurring patterns
• CPC Class Codes: G06V 30/2264 using word shape
• CPC Class Codes: G06V 30/412 Layout analysis of documents structured with printed lines or input boxes, e.g. business forms or tables
• CPC Class Codes: G06V 30/414 Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text
• CPC Class Codes: G06V 30/418 Document matching, e.g. of document images

Abstract

Candidate identification utilizing fingerprint identification is disclosed. The method includes receiving a candidate image comprising a plurality of constituent elements arranged in a content pattern, compensating for rotation variation in the content pattern of the received candidate, analyzing each of the plurality of constituent elements comprising the content pattern of the received candidate image to define a bounded area about each of the plurality of constituent elements, building a candidate fingerprint representative of the content pattern wherein the candidate fingerprint is based on the defined bounded area, comparing the candidate fingerprint to a plurality of fingerprints wherein each of the plurality of fingerprints represents one of a plurality of exemplars, identifying one of the plurality of fingerprints that corresponds to the candidate fingerprint, and evaluating the candidate and one or more identified exemplars to determine the best match there between, wherein the identified exemplar corresponds to the one of the plurality of fingerprints.

Claims

1. A method of candidate identification utilizing fingerprint identification, the method comprising:

receiving a candidate comprising content arranged in a content pattern, wherein the received candidate corresponds to one of a plurality of exemplars;

compensating for rotation variation identified in the content pattern of the received candidate;

analyzing each of a plurality of constituent elements that comprise the content pattern of the received candidate in order to build a fingerprint representative of the content pattern;

comparing the fingerprint to a plurality of fingerprints, wherein each of the plurality of fingerprints corresponding to one of the plurality of exemplars;

evaluating the candidate and one or more identified exemplars to determine the best match there between, wherein the identified exemplar corresponds to the one of the plurality of fingerprints.

2. The method of claim 1, wherein the constituent elements of the content pattern include elements selected from the group consisting of:

character components;

word components and text line components.

3. The method of claim 1, wherein compensating for rotation variation further comprises applying a Hough transformation to the plurality of constituent elements.

4. The method of claim 1, wherein the candidate comprises an electronic document image.

5. The method of claim 4, wherein the content pattern comprises at least one of handwritten information, imaging skew, and image noise.

6. The method of claim 4 further comprising:

identifying multiple exemplars from the plurality of exemplars which may correspond to the fingerprint of the candidate.

7. The method of claim 1 further comprising:

designating a candidate as an unknown if it does not correspond to one of a plurality of exemplars.

8. The method of claim 1, wherein the content pattern represents one or more bounded areas comprising the plurality of constituent elements, and wherein each of the plurality of constituent elements within one of the bounded area represents a neighboring element with respect to the remaining constituent elements within the bounded area.

9. The method of claim 1, wherein the fingerprint comprises at least one list of coordinates corresponding to the constituent elements.

10. The method of claim 9, wherein the at least one list includes a first list corresponding to character components;

a second list corresponding to word components and a third list corresponding to text line components.

11. A method of candidate identification utilizing fingerprint identification, the method comprising:

receiving a candidate image comprising a plurality of constituent elements arranged in a content pattern;

compensating for pattern variation in the content pattern of the received candidate image;

analyzing each of the plurality of constituent elements comprising the content pattern of the received candidate image to define a bounded area about each of the plurality of constituent elements;

building a candidate fingerprint representative of the content pattern, wherein the candidate fingerprint is based on the defined bounded area corresponding to each of the plurality of constituent elements;

comparing the candidate fingerprint to a plurality of fingerprints, wherein each of the plurality of fingerprints represents one of a plurality of exemplars;

evaluating the candidate fingerprint against one or more identified fingerprints to determine the best match there between, wherein each of the identified fingerprints corresponds to the one of the plurality of exemplars; and

identifying one of the plurality of exemplars that corresponds to the candidate fingerprint.

12. The method of claim 11, wherein the candidate image represents an electronic document image.

13. The method of claim 11, wherein the content pattern consists of:

handwritten information, form information, individual characters or an image.

14. The method of claim 11, wherein the constituent elements comprise one or more pixels arranged in a neighborhood.

15. The method of claim 11, wherein the constituent elements of the content pattern include elements selected from the group consisting of:

character components;

word components and text line components.

16. The method of claim 11, wherein compensating for pattern variation comprises compensating for rotation variation.

17. The method of claim 16, wherein compensating for rotation variation further comprises applying a Hough transformation to the plurality of constituent elements.

18. The method of claim 11, wherein identifying one of the plurality of fingerprints includes identifying multiple fingerprints and wherein each of the multiple fingerprints corresponds to one the plurality of exemplars.

19. The method of claim 18 further comprising:

designating the candidate image as an unknown candidate if it does not correspond to one of the plurality of exemplars.

20. A method of identifying a candidate based on candidate content arranged in a content pattern, wherein the candidate corresponds to one of a plurality of exemplars, the method comprising:

correcting for variation in a candidate image by analyzing each of a plurality of constituent elements of the content pattern;

generating a fingerprint representative of the content pattern;

comparing the generated fingerprint to a plurality of fingerprints arranged such that each fingerprint corresponds to one of the plurality of exemplars; and

evaluating the generated fingerprint against one or more identified fingerprints to determine the best match there between, wherein each of the identified fingerprints corresponds to the one of the plurality of exemplars.

21. The method of claim 20, wherein the candidate image is an electronic document image.

22. The method of claim 20, wherein the content pattern comprises at least one of handwritten information, imaging skew, or image noise.

23. The method of claim 20 further comprising:

designating a candidate as an unknown if it does not correspond to one of a plurality of exemplars.

24. The method of claim 23, wherein the plurality of constituent elements are derived from the group consisting of:

text line components, word components and character components.

25. The method of claim 24, wherein the text line components, word components and character components are each represented by a single point.

26. The method of claim 24, wherein generating the fingerprint representative of the content pattern further comprises:

deriving a boundary rectangle for each of the plurality of constituent elements of the content pattern, wherein each boundary rectangle encompasses one of the plurality of constituent elements.

27. The method of claim 26 further comprising:

calculating a single point representative of the derived boundary rectangle.

28. The method of claim 27, wherein correcting for variation comprises correcting for rotation variation.

29. The method of claim 28, wherein correcting for rotation variation further comprises applying a Hough transformation to the single point representative of the derived boundary rectangle.