Method and apparatus to generate object descriptor using extended curvature gabor filter

• US 8,977,040 B2
• Filed: 08/08/2011
• Issued: 03/10/2015
• Est. Priority Date: 09/09/2010
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Hwang, Won Jun, Roh, Kyung Shik, Yoon, Suk June
• Primary Examiner(s): Chu, Randolph I
• Assistant Examiner(s): BLOOM, NATHAN J
• Application Number:
  US13/204,933
• Publication Number:
  US 20120063673A1
• Time in Patent Office: 1,310 Days
• US Class Current: 382/159
• CPC Class Codes: G06F 18/2132 based on discrimination criteria, e.g. discriminant analysis
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 7/42 using transform domain methods
• CPC Class Codes: G06V 10/449 Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
• CPC Class Codes: G06V 10/7715 Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
• CPC Class Codes: G06V 40/168 Feature extraction; Face representation

Abstract

A method and apparatus to generate an object descriptor using extended curvature gabor filters. The method and apparatus may increase a recognition rate of even a relatively small image with use of an extended number of curvature gabor filters having controllable curvatures and may reduce the amount of calculation required for face recognition by performing the face recognition using only some of the extended curvature gabor filters which have a great effect on the recognition rate. The object descriptor generating method includes extracting gabor features from an input object image by applying a plurality of curvature gabor filters, generated via combination of a plurality of curvatures and a plurality of Gaussian magnitudes, to the object image, and generating an object descriptor for object recognition by projecting the extracted features onto a predetermined base vector.

Claims

1. A method of generating an object descriptor, comprising:

extracting, by a computer, gabor features from an input object image by applying a plurality of curvature gabor filters, generated via combination of a plurality of curvatures and a plurality of Gaussian magnitudes so that each curvature gabor filter of the plurality of curvature gabor filters has a different combination of curvature and Gaussian magnitude than the other curvature gabor filters of the plurality of curvature gabor filters, to the object image; and

generating, by a computer, an object descriptor for object recognition by projecting the extracted features onto a predetermined base vector.

2. The method according to claim 1, wherein:

the curvatures are in the range of a value equal to or greater than about 0.0 to a value equal to or less than about 0.2; and

the Gaussian magnitudes are in the range of a value equal to or greater than about 0.5π

to a value equal to or less than about 2π

.

3. The method according to claim 2, wherein:

the curvatures are represented by c={0.0, 0.05, 0.1, 0.2}; and

the Gaussian magnitudes are represented by σ

={0.5π

, π

, 2π

}.

4. The method according to claim 3, wherein, when the plurality of gabor filters are arranged in columns on the basis of the curvatures and in rows on the basis of the Gaussian magnitudes, the gabor features are extracted using some of the gabor filters which are arranged in a T-shaped form.

5. The method according to claim 4, wherein the filters arranged in a T-shaped form include:

filters having the Gaussian magnitude σ

={0.5π

, π

, 2π

} based on the curvature c={0.0}; and

filters having the Gaussian magnitude σ

={π

} based on the curvature c={0.05, 0.1, 0.2}.

6. The method according to claim 1, further comprising:

extracting gabor features from a training object image by applying the curvature gabor filters to the training object image; and

selecting gabor features by performing boosting learning for object image classification with respect to the extracted gabor features, and establishing a gabor feature set including the selected gabor features,

wherein the extraction of the gabor features includes extracting gabor features by applying curvature gabor filters having the established gabor feature set to the input object image.

7. The method according to claim 6, further comprising generating a base vector by performing linear discriminant analysis of the established gabor feature set,

wherein the generation of the object descriptor includes generating an object descriptor for object recognition by projecting the extracted gabor features onto the generated base vector.

8. A method of generating an object descriptor, comprising:

extracting, by a computer, first gabor features from a training object image by applying curvature gabor filters to the training object image;

selecting gabor features by performing boosting learning for object image classification with respect to the extracted gabor features and establishing a gabor feature set including the selected gabor features;

generating, by a computer, a base vector by performing linear discriminant analysis of the established gabor feature set;

extracting, by a computer, second gabor features from an input object image by applying a plurality of curvature gabor filters, generated via combination of a plurality of curvatures and a plurality of Gaussian magnitudes among the established gabor feature set so that each curvature gabor filter of the plurality of curvature gabor filters has a different combination of curvature and Gaussian magnitude than the other curvature gabor filters of the plurality of curvature gabor filters, to the input object image; and

generating, by a computer, an object descriptor for object recognition by projecting the extracted second gabor features onto the generated base vector.

9. The method according to claim 8, wherein:

the curvatures are in the range of a value equal to or greater than about 0.0 to a value equal to or less than about 0.2; and

the Gaussian magnitudes are in the range of a value equal to or greater than about 0.5π

to a value equal to or less than about 2π

.

10. The method according to claim 9, wherein:

the curvatures are represented by c={0.0, 0.05, 0.1, 0.2}; and

the Gaussian magnitudes are represented by σ

={0.5π

, π

, 2π

}.

11. The method according to claim 10, wherein, when the plurality of gabor filters are arranged in columns on the basis of the curvatures and in rows on the basis of the Gaussian magnitudes, the gabor features are extracted using some of the gabor filters which are arranged in a T-shaped form.

12. The method according to claim 11, wherein the filters arranged in a T-shaped form include:

filters having the Gaussian magnitude σ

={0.0, 0.05, 0.1, 0.2} based on the curvature c={0.0}; and

filters having the Gaussian magnitude σ

={π

} based on the curvature c={0.05, 0.1, 0.2}.

13. An apparatus of generating an object descriptor, comprising:

a first feature extractor to extract gabor features from an input object image by applying a plurality of curvature gabor filters, generated via combination of a plurality of curvatures and a plurality of Gaussian magnitudes so that each curvature gabor filter of the plurality of curvature gabor filters has a different combination of curvature and Gaussian magnitude than the other curvature gabor filters of the plurality of curvature gabor filters, to the object image; and

an object descriptor generator to generate an object descriptor for object recognition by projecting the extracted features onto a predetermined base vector.

14. The apparatus according to claim 13, wherein:

the curvatures are in the range of a value equal to or greater than about 0.0 to a value equal to or less than about 0.2; and

the Gaussian magnitudes are in the range of a value equal to or greater than about 0.5π

to a value equal to or less than about 2π

.

15. The apparatus according to claim 14, wherein:

the curvatures are represented by c={0.0, 0.05, 0.1, 0.2}; and

the Gaussian magnitudes are represented by σ

={0.5π

, π

, 2π

}.

16. The apparatus according to claim 15, wherein, when the plurality of gabor filters are arranged in columns on the basis of the curvatures and in rows on the basis of the Gaussian magnitudes, the gabor features are extracted using some of the gabor filters which are arranged in a T-shaped form.

17. The apparatus according to claim 16, wherein the filters arranged in a T-shaped form include:

filters having the Gaussian magnitude σ

={0.5π

, π

, 2π

} based on the curvature c={0.0}; and

filters having the Gaussian magnitude σ

={π

} based on the curvature c={0.05, 0.1, 0.2}.

18. The apparatus according to claim 13, further comprising:

a second feature extractor to extract gabor features from a training object image by applying the curvature gabor filters to the training object image; and

a selector to select gabor features by performing boosting learning for object image classification with respect to the gabor features extracted from the second feature extractor and to establish a gabor feature set including the selected gabor features,

wherein the first feature extractor extracts gabor features by applying curvature gabor filters having the established gabor feature set to the input object image.

19. The apparatus according to claim 18, further comprising generating a base vector generator to generate a base vector by performing linear discriminant analysis of the established gabor feature set,

wherein the object descriptor generator generates an object descriptor for object recognition by projecting the gabor features, extracted from the first feature extractor, onto the generated base vector.