METHODS AND APPARATUS TO PERFORM IMAGE CLASSIFICATION BASED ON PSEUDORANDOM FEATURES

• US 20100272350A1
• Filed: 04/27/2009
• Published: 10/28/2010
• Est. Priority Date: 04/27/2009
• Status: Active Grant

• Current Assignee: The Nielsen Company LLC (MOOD Media Corporation)
• Original Assignee: The Nielsen Company LLC (MOOD Media Corporation)
• Inventors: Lee, Morris
• Granted Patent: US 8,351,712 B2
• Time in Patent Office: Days
• US Class Current: 382/159
• 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 18/241 relating to the classification model, e.g. parametric or non-parametric approaches
• CPC Class Codes: G06V 10/40 Extraction of image or video features
• 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 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 40/16 Human faces, e.g. facial parts, sketches or expressions

Abstract

Example methods and apparatus to perform image classification based on pseudorandom features are disclosed. A disclosed example method includes generating first and second pseudorandom numbers, extracting a first feature of an image based on the first and second pseudorandom numbers, and determining a classification for the image based on the first extracted feature.

Claims

1. A method comprising:

generating first and second pseudorandom numbers;

extracting a first feature of an image based on the first and second pseudorandom numbers; and

determining a classification for the image based on the first extracted feature.

3. A method as defined in claim 1, wherein generating the first pseudorandom number comprises selecting the first pseudorandom numbers from a sequence of numbers approximating a random number property.

5. A method as defined in claim 1, wherein extracting the first feature of the image comprises:

identifying a location of a region of the image based on the first and second pseudorandom numbers; and

computing an average of pixels of the identified, region.

7. A method as defined in claim 1, wherein extracting the first feature of the image comprises:

identifying a location in the image based, on the first and second pseudorandom numbers; and

computing at least one of an edge strength, an edge direction, or a color at the identified location.

8. A method as defined in claim 1, further comprising detecting a region of the image, wherein the first feature is extracted from the detected region.

10. A method as defined in claim 1, further comprising deterministically extracting a second feature of the image, wherein determining the classification is based on the first and second features.

11. A method as defined in claim 10, wherein the second, feature is at least one of a brightness, a color, a position, a feature size, an edge strength, or an edge direction.

15. A method as defined in claim 1, wherein determining the classification for the image comprises:

providing the first extracted feature into a support vector machine; and

receiving the classification from the support vector machine.

17. A method as defined in claim 1, further comprising:

receiving a second, image and a second classification for the second image;

extracting a second feature of the second image based on the first and second pseudorandom numbers; and

training an image classifier based on the second feature and the second classification.

2. (canceled)

4. (canceled)

6. (canceled)

9. (canceled)

12. (canceled)

13. (canceled)

14. (canceled)

16. (canceled)

18. An apparatus comprising:

a pseudorandom number generator to generate one or more pseudorandom numbers;

a feature extractor to extract a first feature of an image based on the one or more pseudorandom numbers; and

a classifier to determine a classification for the image based on the first extracted feature.

20. An apparatus as defined in claim 18, wherein the number generator is to generate one or more additional pseudorandom numbers, and wherein the feature extractor is to identify a region of the image based on the one or more pseudorandom numbers and the one or more additional pseudorandom numbers and to compute an average of pixels of the identified region.

21. An apparatus as defined in claim 18, wherein the first feature comprises at least one of an edge strength, an edge direction, or a color at the identified location.

23. An apparatus as defined in claim 18, further comprising a second feature extractor to deterministically extract a second feature of the image, wherein the classifier is to determine the classification based on the first and second features.

24. An apparatus as defined in claim 23, wherein the second feature is at least one of a brightness, a color, a position, a feature size, an edge strength, or an edge direction.

26. An apparatus as defined in claim 18, wherein the classifier comprises a support vector machine.

27. An apparatus as defined in claim 18, wherein the pseudorandom number generator comprises at least one of a linear congruential generator, a lagged Fibonacci generator, a linear feedback shift register, or a generalized feedback shift register.

19. (canceled)

22. (canceled)

25. (canceled)

28. An article of manufacture storing machine-readable instructions which, when executed, cause a machine to:

generate one or more pseudorandom numbers;

extract a first feature of an image based on the one or more pseudorandom numbers; and

determine a classification for the image based on the first extracted feature.

29. An article of manufacture as defined in claim 28, wherein the machine-readable instructions, when executed, cause the machine to:

generate additional pseudorandom numbers;

identify a region of the image based on the one or more pseudorandom numbers and the additional pseudorandom numbers; and

compute an average of pixels of the identified region.

32. An article of manufacture as defined in claim 28, wherein the machine-readable instructions, when executed, cause the machine to:

receive a second image and a second classification for the second image;

extract a second feature of the second image based on the one or more pseudorandom numbers; and

train an image classifier based, on the second feature and the second, classification.

33. An article of manufacture as defined in claim 28, wherein the first feature of the image comprises at least one of an edge strength, an edge direction, or a color at the identified location.

30. (canceled)

31. (canceled)