Methods and systems for identifying digital image characteristics

• US 8,160,365 B2
• Filed: 06/30/2008
• Issued: 04/17/2012
• Est. Priority Date: 06/30/2008
• Status: Active Grant

• Current Assignee: Sharp Kabushiki Kaisha (Hon Hai Precision Industry Co., Ltd.)
• Original Assignee: Sharp Laboratories of America Incorporated (Hon Hai Precision Industry Co., Ltd.)
• Inventors: Campbell, Richard John, Speigle, Jon M.
• Primary Examiner(s): MARIAM, DANIEL G
• Application Number:
  US12/164,985
• Publication Number:
  US 20090324083A1
• Time in Patent Office: 1,387 Days
• Field of Search: 382/182, 382185-187, 382/190, 382/203, 382/228
• US Class Current: 382/190
• CPC Class Codes: G06F 18/24155 Bayesian classification
• CPC Class Codes: G06V 10/242 by image rotation, e.g. by 90 degrees
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects

Abstract

Aspects of the present invention relate to methods and systems for determining image characteristics in a digital image.

Claims

1. A method for determining an image characteristic in a digital image, said method comprising:

a) identifying a first glyph component in a digital image;

b) determining a first reference location for said first glyph component;

c) in said first glyph component, determining a first feature location associated with a first glyph feature;

d) determining a first position, relative to said first reference location, of said first feature location;

e) accumulating, in a first distribution counter, a frequency-of-occurrence count related to said first glyph feature and said first position;

f) determining a distribution feature based on said first distribution counter; and

g) associating a first image-characteristic value with said digital image based on said distribution feature and a first probabilistic model, wherein said first probabilistic model is associated with said first glyph feature and said first image characteristic.

2. A method according to claim 1 further comprising receiving a first plurality of parameters associated with said first probabilistic model.

3. A method according to claim 1, wherein said identifying comprises forming a binary image from said digital image.

4. A method according to claim 1 further comprising associating a second image-characteristic value with said digital image based on said distribution feature and a second probabilistic model, wherein said second probabilistic model is associated with said first glyph features and said second image characteristic.

5. A method according to claim 4, wherein:

a) said first image characteristic is an orientation characteristic; and

b) said second image characteristic is a language characteristic.

6. A method according to claim 4, wherein:

a) said first probabilistic model and said second probabilistic model are a joint probabilistic model; and

b) said associating a first image-characteristic value and said associating a second image-characteristic value is done jointly according to said joint probabilistic model.

7. A method according to claim 1, wherein said associating comprises permuting said distribution feature to correspond to a first orientation, wherein said first glyph component is associated with a second orientation that relates to said first orientation by a rotation.

8. A method according to claim 1, wherein said probabilistic model comprises a normal density function.

9. A method according to claim 1, wherein said associating comprises a classifier selected from the group consisting of a maximum a posteriori probability (MAP) classifier, a maximum likelihood (ML) classifier and a minimum-risk classifier.

10. A method according to claim 1, wherein said first reference location is the centroid of said first glyph component.

11. A method according to claim 1, wherein said determining a first reference location comprises:

a) determining a bounding box for said first glyph component; and

b) determining the center of said bounding box.

12. A method according to claim 1, wherein said determining a first position comprises a quadrant decomposition based on said first reference location.

13. A method according to claim 1, wherein said first image characteristic is a characteristic selected from the group consisting of an orientation characteristic and a language characteristic.

14. A method according to claim 1 further comprising determining a confidence level associated with said first image characteristic.

15. A method according to claim 1 further comprising:

a) in said first glyph component, determining a second feature location associated with a second glyph feature;

b) determining a second position, relative to said first reference location, of said second feature location; and

c) accumulating, in said first distribution counter, a frequency-of-occurrence count related to said second glyph feature and said second position.

16. A method according to claim 1 further comprising:

a) determining a second glyph component in said digital image;

b) determining a second reference location for said second glyph component;

c) in said second glyph component, determining a second feature location associated with said first glyph feature;

d) determining a second position, relative to said second reference location, of said second feature location;

e) accumulating, in a second distribution counter, a frequency-of-occurrence count related to said first glyph feature and said second position; and

f) combining said second distribution counter and said first distribution counter to form said distribution feature.

17. A method for determining an image characteristic in a digital image, said method comprising:

a) identifying a first plurality of glyph components in a digital image;

b) for each glyph component in said first plurality of glyph components, determining a glyph-feature location-distribution counter, thereby producing a first plurality of distribution counters;

c) determining a distribution feature based on said first plurality of distribution counters; and

d) associating a first image-characteristic value with said digital image based on said distribution feature and a first probabilistic model, wherein said first probabilistic model is associated with a plurality of glyph features and said first image characteristic.

18. A method according to claim 17, wherein said associating comprises a classifier selected from the group consisting of a maximum a posteriori probability (MAP) classifier, a maximum likelihood (ML) classifier and a minimum-risk classifier.

19. A method according to claim 17, wherein said first image characteristic is a characteristic selected from the group consisting of an orientation characteristic and a language characteristic.

20. A method according to claim 17, wherein said identifying a first plurality of glyph components comprises determining likely text glyph components from a first plurality of connected components.

21. A method according to claim 17, wherein said identifying a first plurality of glyph components comprises filtering a second plurality of glyph components to eliminate noise.

22. A method for determining an image characteristic in a digital image, said method comprising:

a) identifying a first glyph component in a digital image;

b) identifying a first pattern associated with a first glyph pixel within said first glyph component, wherein said first pattern is based on said first glyph pixel and a first plurality of pixels within a neighborhood proximate to said first glyph pixel;

c) accumulating, in a first distribution counter, a frequency-of-occurrence count related to said first pattern;

d) determining a distribution feature based on said first distribution counter; and

e) associating a first image-characteristic value with said digital image based on said distribution feature and a first probabilistic model, wherein said first probabilistic model is associated with said first pattern and said first image characteristic.

23. A method according to claim 22, wherein said first image characteristic is a characteristic selected from the group consisting of an orientation characteristic and a language characteristic.