Face detection using division-generated Haar-like features for illumination invariance

• US 8,977,056 B2
• Filed: 10/31/2013
• Issued: 03/10/2015
• Est. Priority Date: 07/26/2010
• Status: Active Grant

• Current Assignee: FotoNation Limited (Adeia, Inc.)
• Original Assignee: FotoNation Limited (Adeia, Inc.)
• Inventors: Susanu, George, Filip, Dan, Gangea, Mihnea
• Primary Examiner(s): Patel, Jayesh A
• Assistant Examiner(s): Roz, Mark
• Application Number:
  US14/069,097
• Publication Number:
  US 20140056527A1
• Time in Patent Office: 495 Days
• Field of Search: None
• US Class Current: 382/181
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

Faces in images are quickly detected with minimal memory resource usage. Instead of calculating a Haar-like feature value by subtracting the average pixel intensity value in one rectangular region from the average pixel intensity value in another, adjacent rectangular region, a face-detection system calculates that Haar-like feature value by dividing the average pixel intensity value in one rectangular region by the average pixel intensity value in another adjacent rectangular region. Thus, each Haar-like value is calculated as a ratio of average pixel intensity values rather than as a difference between such average pixel intensity values. The feature values are calculated using this ratio-based technique both during the machine-learning procedure, in which the numerical ranges for features in known face-containing images are learned based on labeled training data, and during the classifier-applying procedure, in which an unlabeled image'"'"'s feature values are calculated and compared to the previously machine-learned numerical ranges.

Claims

1. A method for determining whether a portion of an image contains a face, comprising:

for each region of an N number of regions of an image, determining a region value for that region such that an N number of region values are derived, wherein the N regions are within a particular portion of the image, and wherein N is an integer greater than two;

generating a first feature value for a first feature, wherein generating the first feature value comprises;

determining a first ratio by dividing a first sum of a first subset of the N number of region values by a second sum of a second subset of the N number of region values, wherein the first subset of the N number of region values corresponds to different regions than the second subset of the N number of region values; and

determining the first feature value based on the first ratio;

obtaining a first value range that is associated with the first feature, wherein first feature values within the first value range indicate a likelihood that a face is present;

determining whether the first feature value falls within the first value range; and

in response to determining that the first feature value does not fall within the first value range, storing data that indicates that the particular portion of the image does not contain a face;

wherein the method is performed by one or more computing devices.

2. The method of claim 1,

wherein the N number of regions comprises a first region, a second region adjacent to the first region, and a third region adjacent to the second region;

wherein a first region value is determined for the first region, a second region value is determined for the second region, and a third region value is determined for the third region; and

wherein the first ratio is the ratio between the second region value and the sum of the first and third region values.

3. The method of claim 2, wherein the first region value is a first pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the first region, wherein the second region value is a second pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the second region, and wherein the third region value is a third pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the third region.

4. The method of claim 1,

wherein the N number of regions comprises a first region, a second region adjacent to the first region, a third region adjacent to the first region, and a fourth region adjacent to the second and third regions;

wherein a first region value is determined for the first region, a second region value is determined for the second region, a third region value is determined for the third region, and a fourth region value is determined for the fourth region; and

wherein the first ratio is the ratio between the sum of the first and fourth region values and the sum of the second and third region values.

5. The method of claim 4, wherein the first region value is a first pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the first region, wherein the second region value is a second pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the second region, wherein the third region value is a third pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the third region, and wherein the fourth region value is a fourth pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the fourth region.

6. The method of claim 1, wherein the first feature is a Haar-like feature.

7. The method of claim 1, wherein the image is a source image, and wherein the first feature value is generated without using an integral image generated from the source image.

8. The method of claim 1, wherein the first feature value is generated without using a statistical variance parameter that reflects variance of pixel intensity values in the image.

9. A non-transitory computer readable storage medium storing instructions which, when executed by one or more processors, cause the one or more processors to perform operations comprising:

for each region of an N number of regions of an image, determining a region value for that region such that an N number of region values are derived, wherein the N regions are within a particular portion of the image, and wherein N is an integer greater than two;

generating a first feature value for a first feature, wherein generating the first feature value comprises;

determining a first ratio by dividing a first sum of a first subset of the N number of region values by a second sum of a second subset of the N number of region values, wherein the first subset of the N number of region values corresponds to different regions than the second subset of the N number of region values; and

determining the first feature value based on the first ratio;

obtaining a first value range that is associated with the first feature, wherein first feature values within the first value range indicate a likelihood that a face is present;

determining whether the first feature value falls within the first value range; and

in response to determining that the first feature value does not fall within the first value range, storing data that indicates that the particular portion of the image does not contain a face.

10. The computer readable storage medium of claim 9,

wherein the N number of regions comprises a first region, a second region adjacent to the first region, and a third region adjacent to the second region;

wherein a first region value is determined for the first region, a second region value is determined for the second region, and a third region value is determined for the third region; and

wherein the first ratio is the ratio between the second region value and the sum of the first and third region values.

11. The computer readable storage medium of claim 10, wherein the first region value is a first pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the first region, wherein the second region value is a second pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the second region, and wherein the third region value is a third pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the third region.

12. The computer readable storage medium of claim 9,

wherein the N number of regions comprises a first region, a second region adjacent to the first region, a third region adjacent to the first region, and a fourth region adjacent to the second and third regions;

wherein a first region value is determined for the first region, a second region value is determined for the second region, a third region value is determined for the third region, and a fourth region value is determined for the fourth region; and

wherein the first ratio is the ratio between the sum of the first and fourth region values and the sum of the second and third region values.

13. The computer readable storage medium of claim 12, wherein the first region value is a first pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the first region, wherein the second region value is a second pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the second region, wherein the third region value is a third pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the third region, and wherein the fourth region value is a fourth pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the fourth region.

14. The computer readable storage medium of claim 9, wherein the first feature is a Haar-like feature.

15. The computer readable storage medium of claim 9, wherein the image is a source image, and wherein the first feature value is generated without using an integral image generated from the source image.

16. The computer readable storage medium of claim 9, wherein the first feature value is generated without using a statistical variance parameter that reflects variance of pixel intensity values in the image.

17. A device comprising:

one more storages capable of storing one or more images and other data; and

a face detector configured to perform operations comprising;

for each region of an N number of regions of an image, determining a region value for that region such that an N number of region values are derived, wherein the N regions are within a particular portion of the image, and wherein N is an integer greater than two;

generating a first feature value for a first feature, wherein generating the first feature value comprises;

determining a first ratio by dividing a first sum of a first subset of the N number of region values by a second sum of a second subset of the N number of region values, wherein the first subset of the N number of region values corresponds to different regions than the second subset of the N number of region values;

determining the first feature value based on the first ratio;

obtaining a first value range that is associated with the first feature, wherein first feature values within the first value range indicate a likelihood that a face is present;

determining whether the first feature value falls within the first value range; and

in response to determining that the first feature value does not fall within the first value range, storing data that indicates that the particular portion of the image does not contain a face.

18. The device of claim 17,

wherein the N number of regions comprises a first region, a second region adjacent to the first region, and a third region adjacent to the second region;

wherein a first region value is determined for the first region, a second region value is determined for the second region, and a third region value is determined for the third region; and

wherein the first ratio is the ratio between the second region value and the sum of the first and third region values.

19. The device of claim 18, wherein the first region value is a first pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the first region, wherein the second region value is a second pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the second region, and wherein the third region value is a third pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the third region.

20. The device of claim 17,

wherein the N number of regions comprises a first region, a second region adjacent to the first region, a third region adjacent to the first region, and a fourth region adjacent to the second and third regions;

wherein a first region value is determined for the first region, a second region value is determined for the second region, a third region value is determined for the third region, and a fourth region value is determined for the fourth region; and

wherein the first ratio is the ratio between the sum of the first and fourth region values and the sum of the second and third region values.

21. The device of claim 20, wherein the first region value is a first pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the first region, wherein the second region value is a second pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the second region, wherein the third region value is a third pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the third region, and wherein the fourth region value is a fourth pixel intensity value that is derived based upon pixel intensity values of two or more pixels within the fourth region.

22. The device of claim 17, wherein the first feature is a Haar-like feature.

23. The device of claim 17, wherein the image is a source image, and wherein the first feature value is generated without using an integral image generated from the source image.

24. The device of claim 17, wherein the first feature value is generated without using a statistical variance parameter that reflects variance of pixel intensity values in the image.