Method for determining a correlation between images using multi-element image descriptors

• US 20010051004A1
• Filed: 02/23/2001
• Published: 12/13/2001
• Est. Priority Date: 10/19/1998
• Status: Active Grant

• Current Assignee: Lizhi Wang
• Original Assignee: Lizhi Wang
• Inventors: Wang, Lizhi
• Granted Patent: US 6,687,416 B2
• Time in Patent Office: Days
• US Class Current: 382/206
• CPC Class Codes: G06T 7/45 using co-occurrence matrix computation
• CPC Class Codes: G06V 10/42 Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
• CPC Class Codes: G06V 10/7515 Shifting the patterns to accommodate for positional errors
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

A method of determining a correlation between images, wherein each image is represented by an image descriptor. Each image descriptor comprises a peak descriptor, a set of statistic descriptors, and a set of co-occurence descriptors. Initially, a category associated with each image descriptor is determined. Provided the image descriptors belong to a common category, a difference value between select corresponding components of respective image descriptors is determined in accordance with an associated correlation protocol, the difference value representing the correlation between the images represented by respective image descriptors.

Claims

1. A method of generating an image descriptor for use in image processing, the method comprising:

filtering image data to provide orientation data and energy data;

generating a first set of global information and local information for the orientation data and a second set of global information and local information for the energy data;

extracting descriptors from the respective sets of global information and local information associated with the orientation data and the energy data;

combining the descriptors associated with the orientation data and the descriptors associated with the energy data to form an image descriptor.

2. The method of

claim 1

, wherein the step of generating a first set of global information and local information for the orientation data comprises:

generating a histogram for each information band of the image data; and

generating a co-ocurrence matrix corresponding to the image data.

3. The method of

claim 2

, wherein the descriptors extracted from the first set of global information and local information associated with the orientation data comprises peak descriptor data, statistic descriptor data, and co-occurence descriptor data.

4. The method of

claim 1

, wherein the step of generating a second set of global information and local information for the energy data comprises:

generating a histogram for each information band of the image data; and

generating a co-ocurrence matrix corresponding to the image data.

5. The method of

claim 4

, wherein the descriptors extracted from the second set of global information and local information associated with the energy data comprises statistic descriptor data and co-occurence descriptor data.

6. A method of determining a correlation between images, wherein each image is represented by an image descriptor, each image descriptor comprising a peak descriptor, a set of statistic descriptors, and a set of co-occurence descriptors, the method comprising:

determining a category associated with each image descriptor; and

determining a difference value between select corresponding elements of respective image descriptors in accordance with an associated correlation protocol provided the image descriptors belong to a common category, the difference value representing the correlation between the images represented by respective image descriptors.

7. The method of

claim 6

, wherein the images comprise multi-band images, each multi-band image having an associated image descriptor for each information band of the multi-band image.

8. The method of

claim 6

, wherein the set of statistic descriptors comprise an orientation statistic descriptor and an energy statistic descriptor.

9. The method of

claim 6

, wherein the set of co-occurence matrix descriptors comprise an orientation co-occurence descriptor and an energy co-occurence descriptor.

10. The method of

claim 6

, wherein the peak descriptor contains data indicating a category type associated with a corresponding image descriptor.

11. The method of

claim 10

, wherein the image descriptors correspond to a large peak category type.

12. The method of

claim 11

, further including the step of determining a correlation between corresponding peak information contained in the respective image descriptors.

13. The method of

claim 12

, wherein the step of determining a difference value between the image descriptors comprises:

determining a difference between a corresponding set of statistic descriptors associated with the respective image descriptors;

determining a difference between a corresponding peak descriptors associated with the respective image descriptors; and

determining a difference between a corresponding set of co-occurence matrix descriptors associated with the respective image descriptors.

14. The method of

claim 13

, further including the step of comparing the difference value to a correlation value to determine a magnitude of correlation between the image descriptors.

15. The method of

claim 10

, wherein the image descriptors correspond to a smooth category type.

16. The method of

claim 15

, wherein the step of determining a difference value between the image descriptors comprises:

determining a difference between a corresponding set of statistic descriptors associated with the respective image descriptors; and

determining a difference between a corresponding set of co-occurence matrix descriptors associated with the respective image descriptors.

17. The method of

claim 16

, further including the step of comparing the difference value to a correlation value to determine a magnitude of correlation between the image descriptors.

18. The method of

claim 10

, wherein the image descriptors correspond to a non-orientation category type.

19. The method of

claim 18

, wherein the step of determining a difference value between the image descriptors comprises:

classifying the orientation of each image descriptor for each information band of a respective image;

removing non-orientation peak information from each image descriptor associated with the respective images provided each image descriptor for each information band corresponds to a non-orientation classification; and

re-processing the image descriptors associated with each respective image.

20. The method of

claim 18

, where the step of determining a difference value between the image descriptors comprises:

classifying the orientation of each image descriptor for each information band of a respective image;

removing image descriptors associated with information bands of a multi-band image provided that the image descriptors being removed image fail to correspond to a non-orientation classification; and

re-processing each remaining image descriptor.

21. A computer software product including a medium readable by a processor, the medium having stored thereon a sequence of executable instructions which, when executed by a processor, cause the processor to:

determine a category associated with image descriptors, each image descriptor comprising a peak descriptor, a set of statistic descriptors, and a set of co-occurence descriptors; and

determine a difference value between select corresponding components of respective image descriptors in accordance with an associated correlation protocol provided the image descriptors belong to a common category, the difference value representing a correlation between images represented by respective image descriptors.

22. The computer software product of

claim 21

, wherein the images comprise multi-band images, each multi-band image having an associated image descriptor for each information band of the multi-band image.

23. The computer software product of

claim 21

, wherein the set of statistic descriptors comprise an orientation statistic descriptor and an energy statistic descriptor.

24. The computer software product of

claim 21

, wherein the set of co-occurence matrix descriptors comprise an orientation co-occurence descriptor and an energy co-occurence descriptor.

25. The computer software product of

claim 21

, wherein the peak descriptor contains data indicating a category type associated with a corresponding image descriptor.

26. The computer software product of

claim 25

, wherein the image descriptors correspond to a large peak category type.

27. The computer software product of

claim 26

, further including the step of determining a correlation between corresponding peak information contained in the respective image descriptors.

28. The computer software product of

claim 27

, wherein the step of determining a difference value between the image descriptors comprises:

determining a difference between a corresponding set of statistic descriptors associated with the respective image descriptors;

determining a difference between a corresponding peak descriptors associated-with the respective image descriptors; and

determining a difference between a corresponding set of co-occurence matrix descriptors associated with the respective image descriptors.

29. The computer software product of

claim 28

, further including the step of comparing the difference value to a correlation value to determine a magnitude of correlation between the image descriptors.

30. The computer software product of

claim 25

, wherein the image descriptors correspond to a smooth category type.

31. The computer software product of

claim 30

, wherein the step of determining a difference value between the image descriptors comprises:

determining a difference between a corresponding set of statistic descriptors associated with the respective image descriptors; and

determining a difference between a corresponding set of co-occurence matrix descriptors associated with the respective image descriptors.

32. The computer software product of

claim 31

, further including the step of comparing the difference value to a correlation value to determine a magnitude of correlation between the image descriptors.

33. The computer software product of

claim 25

, wherein the image descriptors correspond to a non-orientation category type.

34. The computer software product of

claim 33

, wherein the step of determining a difference value between the image descriptors comprises:

classifying the orientation of each image descriptor for each information band of a respective image;

removing non-orientation peak information from each image descriptor associated with the respective images provided each image descriptor for each information band corresponds to a non-orientation classification; and

re-processing the image descriptors associated with each respective image.

35. The computer software product of

claim 33

, where the step of determining a difference value between the image descriptors comprises:

classifying the orientation of each image descriptor for each information band of a respective image;

removing image descriptors associated with information bands of a multi-band image provided that the image descriptors being removed image fail to correspond to a non-orientation classification; and

re-processing each remaining image descriptor.