Method and apparatus for obtaining and processing Gabor image features

• US 8,509,538 B2
• Filed: 12/20/2007
• Issued: 08/13/2013
• Est. Priority Date: 12/20/2007
• Status: Expired due to Fees

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Li, Jianguo, Wang, Tao, Zhang, Yimin
• Primary Examiner(s): Rush, Eric
• Application Number:
  US11/961,649
• Publication Number:
  US 20090161967A1
• Time in Patent Office: 2,063 Days
• Field of Search: 382/100, 382/115, 382/155, 382/159, 382/181, 382/190, 382/195, 382/203, 382/205, 382/210, 382/224, 382/225
• US Class Current: 382/190
• CPC Class Codes: G06F 18/21 Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
• CPC Class Codes: G06F 18/217 Validation; Performance evaluation; Active pattern learning techniques
• CPC Class Codes: G06V 10/449 Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
• CPC Class Codes: G06V 10/70 using pattern recognition or machine learning optical pattern recognition or electronic computations therefor G06V10/88
• CPC Class Codes: G06V 10/776 Validation; Performance evaluation
• CPC Class Codes: G06V 40/165 using facial parts and geometric relationships

Abstract

Machine-readable media, methods, apparatus and system for obtaining and processing image features are described. In some embodiments, a Gabor representation of an image may be obtained by using a Gabor filter. A region may be determined from the Gabor representation, wherein the region comprises a plurality of Gabor pixels of the Gabor representation; and, a sub-region may be determined from the region, wherein the sub-region comprises more than one of the plurality of Gabor pixels. Then, a Gabor feature may be calculated based upon a magnitude calculation related to the sub-region and the region.

Claims

1. A method of obtaining a refined Gabor feature corresponding to a sub-region of an image, comprising:

receiving an image;

obtaining a Gabor representation of the received image by using a Gabor filter;

determining a region from the Gabor representation, wherein the region comprises a first plurality of Gabor pixels of the Gabor representation;

determining a sub-region from the region, wherein the sub-region comprises a second plurality of Gabor pixels of the Gabor representation, each of the Gabor pixels of the sub-region also being Gabor pixels of the region;

calculating a magnitude sum in the Gabor pixels of the region and in the Gabor pixels in the sub-region; and

calculating a refined Gabor feature for the sub-region based upon a magnitude calculation by dividing the magnitude sum for the sub-region by the magnitude sum for and the region.

2. The method of claim 1, further comprising obtaining an integral image from the Gabor representation, wherein an integral pixel of the integral image corresponds to a Gabor pixel of the Gabor representation and a value of the integral pixel is obtained by integrating magnitudes for Gabor pixels within a field defined by the Gabor pixel corresponding to the integral pixel.

3. The method of claim 2, wherein calculating a magnitude sum comprises calculating a magnitude sum based upon the integral image.

4. The method of claim 1, wherein the sub-region is determined with reference to one or more Haar patterns.

5. The method of claim 1, wherein the sub-region is determined with reference to one or more extended Haar patterns.

6. The method of claim 1, wherein the calculating a sub-region magnitude sum comprises:

calculating a first magnitude sum for a first integral pixel;

calculating a second magnitude sum for a second integral pixel, each integral pixel corresponding to a sum of the Gabor pixels within a region defined by an origin and a position corresponding to the integral pixel; and

obtaining the sub-region magnitude sum by subtracting the first magnitude sum from the second magnitude sum.

7. The method of claim 1 wherein obtaining a Gabor representation comprises convolving the image with a two-dimensional Gabor filter.

8. The method of claim 1, wherein the calculating a refined Gabor feature further comprises:

calculating a first magnitude sum for Gabor pixels within the sub-region;

calculating a second magnitude sum for Gabor pixels within a part of the region that does not overlap with the sub-region; and

obtaining a refined Gabor feature by subtracting the second magnitude sum from the first magnitude sum.

9. The method of claim 1, further comprising:

combining the calculated refined Gabor feature with a plurality of other calculated Gabor features to form a training image in which the Gabor features serve as training features;

training image classifiers using the training image.

10. An apparatus, comprising:

a Gabor filter to obtain a Gabor representation of an image;

a region logic to determine a region from the Gabor representation, wherein the region comprises a plurality of Gabor pixels of the Gabor representation;

a sub-region logic to determine a sub-region from the region, wherein the sub-region comprises a second plurality of Gabor pixels of the Gabor representation, each of the Gabor pixels of the sub-region also being Gabor pixels of the region; and

a feature calculating logic to calculate a magnitude sum in the Gabor pixels of the region and in the Gabor pixels in the sub-region and to calculate a refined Gabor feature for the sub-region based upon a magnitude calculation by dividing the magnitude sum for the sub-region by the magnitude sum for the region.

11. The apparatus of claim 10, further comprising an integral logic to obtain an integral image from the Gabor representation, wherein an integral pixel of the integral image corresponds to a Gabor pixel of the Gabor representation, and a value of the integral pixel is obtained by integrating magnitudes for Gabor pixels within a field defined by the Gabor pixel corresponding to the integral pixel.

12. The apparatus of claim 11, wherein the feature calculating logic is further to perform the magnitude sum calculation for the sub-region and the region based upon the integral image.

13. The apparatus of claim 10, wherein the sub-region is determined with reference to one or more Haar patterns.

14. The apparatus of claim 10, wherein the sub-region is determined with reference to one or more extended Haar patterns.

15. The apparatus of claim 10, wherein the feature calculating logic is further to:

calculate a first magnitude sum for a first integral pixel;

calculate a second magnitude sum for a second integral pixel, each integral pixel corresponding to a sum of the Gabor pixels within a region defined by an origin and a position corresponding to the integral pixel; and

obtain the sub-region magnitude sum by subtracting the first magnitude sum from the second magnitude sum.

16. The apparatus of claim 10, wherein the feature calculating logic is further to:

calculate a first magnitude sum for Gabor pixels within the sub-region;

calculate a second magnitude sum for Gabor pixels within a part of the region that does not overlap with the sub-region; and

obtain a refined Gabor feature by subtracting the second magnitude sum from the first magnitude sum.

17. A system, comprising:

a processor;

a machine-readable medium coupled to the processor, wherein the machine-readable medium further comprises a plurality of instructions which, if executed by the processor, cause the system to perform operations comprising;

receiving an image;

obtaining a Gabor representation of the received image by using a Gabor filter;

determining a region from the Gabor representation, wherein the region comprises a first plurality of Gabor pixels of the Gabor representation;

determining a sub-region from the region, wherein the sub-region comprises a second plurality of Gabor pixels of the Gabor representation, each of the Gabor pixels of the sub-region also being Gabor pixels of the region;

calculating a magnitude sum in the Gabor pixels of the region and in the Gabor pixels in the sub-region; and

calculating a refined Gabor feature for the sub-region based upon a magnitude calculation by dividing the magnitude sum for the sub-region by the magnitude sum for the region.

18. The system of claim 17, wherein the plurality of instructions further cause the system to perform operations comprising:

obtaining an integral image from the Gabor representation, wherein an integral pixel of the integral image corresponds to a Gabor pixel of the Gabor representation, and a value of the integral pixel is obtained by integrating magnitudes for Gabor pixels within a field defined by the Gabor pixel corresponding to the integral pixel.

19. The system of claim 18, wherein calculating a magnitude sum comprises calculating a magnitude sum based upon the integral image.

20. The system of claim 17, wherein the sub-region is determined with reference to one or more Haar patterns.

21. The system of claim 17, wherein the sub-region is determined with reference to one or more extended Haar patterns.

22. The system of claim 17, wherein the instructions calculating the sub-region magnitude sum, further cause the system to perform operations comprising:

calculating a first magnitude sum for a first integral pixel;

calculating a second magnitude sum for a second integral pixel, each integral pixel corresponding to a sum of the Gabor pixels within a region defined by an origin and a position corresponding to the integral pixel; and

obtaining the sub-region magnitude sum by subtracting the first magnitude sum from the second magnitude sum.