Method and system for face detection in digital images

• US 20040179719A1
• Filed: 03/12/2003
• Published: 09/16/2004
• Est. Priority Date: 03/12/2003
• Status: Active Grant

• Current Assignee: Monument Peak Ventures, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: Eastman Kodak Company
• Inventors: Chen, Shoupu, Nicponski, Henry, Ray, Lawrence
• Granted Patent: US 7,508,961 B2
• Time in Patent Office: Days
• US Class Current: 382/118
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06F 18/254 of classification results, e.g. of results related to same input data
• CPC Class Codes: G06V 10/446 using Haar-like filters, e.g. using integral image techniques
• CPC Class Codes: G06V 10/774 Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06V 10/809 of classification results, e.g. where the classifiers operate on the same input data
• CPC Class Codes: G06V 40/162 using pixel segmentation or colour matching

Abstract

A digital image processing method for detecting faces in a digital color image comprises the steps of: providing a distributed face detection system having complementary classifiers, wherein the classifiers are complementary in a frequency domain; selecting classifier parameters for the complementary classifiers from a plurality of different parameter generating sources, at least one of which is controllable by human input; reconfiguring the complementary classifiers in the distributed face detection system according to the selected classifier parameters; and detecting faces using the distributed face detection system.

Claims

1. A digital image processing method for detecting faces in a digital color image, said method comprising the steps of:

providing a distributed face detection system having complementary classifiers, wherein the classifiers are complementary in a frequency domain;

selecting classifier parameters for the complementary classifiers from a plurality of different parameter generating sources, at least one of which is controllable by human input;

reconfiguring the complementary classifiers in the distributed face detection system according to the selected classifier parameters; and

detecting faces using the distributed face detection system.

2. The method as claimed in claim 1 wherein the step of providing a distributed face detection system having complementary classifiers comprises the steps of:

constructing one or more classifiers having high execution speed and featuring operation in one portion of the frequency spectrum of the digital image; and

constructing one or more classifiers having high accuracy and featuring operation in another portion of the frequency spectrum of the digital image.

3. The method as claimed in claim 2 wherein the step of constructing one or more classifiers having high execution speed and featuring operation in one portion of the frequency spectrum of the digital image comprises the step of constructing a grid pattern classifier.

4. The method as claimed in claim 3 wherein the step of constructing a grid pattern classifier comprises the steps of:

generating a mean grid pattern element (MGPe) image from a plurality of sample face images;

generating an integral image from the digital color image;

locating faces in the digital color image by using the integral image to perform a correlation test between the mean grid pattern element (MGPe) image and the digital color image at a plurality of effective resolutions by reducing the digital color image to a plurality of grid pattern element images (GPes) at different effective resolutions and correlating the MGPe with the GPes.

5. The method as claimed in claim 4 wherein the step of generating a mean grid pattern element (MGPe) image comprises the steps of:

collecting sample face images;

generating a mean face image from the sample face images;

selecting a grid pattern (GP); and

reducing the resolution of the mean face image to the resolution of the selected grid pattern (GP) by averaging.

6. The method as claimed in claim 5 wherein the grid pattern is regular.

7. The method as claimed in claim 5 wherein the grid pattern is irregular.

8. The method as claimed in claim 7 wherein the step of selecting an irregular grid pattern comprises determining a plurality of different size grid cells that cover major features including eyes, nose, mouth, forehead, and cheek of the mean face image.

9. The method as claimed in claim 6 wherein the step of selecting a regular grid pattern comprises computing a distance e_l between two eye centers of the mean face image;

computing a center position c between two eye centers;

using e_l and c to determine a region that contains M by N grid cells with each cell having m by n pixels.

10. The method as claimed in claim 4 wherein the step of generating an integral image further comprises the steps of:

replacing non-skin-color pixels with black to produce an image C having skin color pixels;

replacing non-face-shaped clusters with black to produce an image E having skin colored face shaped clusters;

labeling clusters of skin colored face shaped clusters; and

generating the integral image from each cluster of the image E.

11. The method claimed in claim 10 further comprising the steps of:

eliminating faces that contain more than a predetermined percentage of black pixels; and

merging faces that substantially overlap.

12. The method claimed in claim 10 wherein the step of replacing non-face-shaped clusters comprises the steps of:

clustering skin-color pixels in image C into clusters;

applying a morphological opening and closing processes to skin-colored pixel clusters; and

replacing the pixels of a cluster with black if it does not meet a geometrical criterion for a face, resulting in an image E, and wherein the step of labeling skin-colored clusters comprises the step of generating a linked list of sets of parameters including a starting position, width, and height that defines regions containing a cluster of skin-colored pixels.

13. The method as claimed in claim 1 wherein the step of selecting classifier parameters for the complementary classifiers from a plurality of different parameter generating sources comprises the steps of:

providing a constant parameter generator; and

providing a controllable parameter generator.

14. The method as claimed in claim 2 wherein the step of constructing one or more classifiers having high accuracy and featuring operation in another portion of the frequency spectrum of the digital image comprises steps of:

constructing a cascaded classifier; and

constructing a Bayesian classifier.

15. The method as claimed in claim 1 wherein the step of reconfiguring the complementary classifiers in the distributed face detection system according to the selected classifier parameters comprises the steps of

providing commands to individual classifiers;

and

directing the input digital image to different classifiers in accordance with the commands.

16. A computer storage medium having instructions stored therein for causing a computer to perform the method of claim 1.

20. The method as claimed in claim 2 wherein the step of constructing one or more classifiers having high accuracy and featuring operation in another portion of the frequency spectrum of the digital image comprises steps of:

constructing a cascaded classifier; and

constructing a Bayesian classifier.

17. A digital image processing method for detecting faces in a digital color image, said method comprising the steps of:

providing a distributed face detection system having complementary classifiers, wherein the classifiers are complementary in a frequency domain and comprise one or more classifiers having high execution speed and featuring operation in one portion of the frequency spectrum of the digital image, and one or more classifiers having high accuracy and featuring operation in another portion of the frequency spectrum of the digital image;

providing classifier parameters for the complementary classifiers from a plurality of different parameter generating sources dependent upon conditions of the digital image; and

detecting faces using the distributed face detection system.

18. The method as claimed in claim 17 wherein the step of constructing one or more classifiers having high execution speed and featuring operation in one portion of the frequency spectrum of the digital image comprises the step of constructing a grid pattern classifier.

19. The method as claimed in claim 17 wherein the step of selecting classifier parameters for the complementary classifiers from a plurality of different parameter generating sources comprises the steps of:

providing a constant parameter generator; and

providing a controllable parameter generator.

21. A computer storage medium having instructions stored therein for causing a computer to perform the method of claim 17.

22. A digital image processing system for detecting faces in a digital color image, said system comprising:

a distributed face detection system having complementary classifiers, wherein the classifiers are complementary in a frequency domain;

a plurality of different parameter generating sources for selecting classifier parameters for the complementary classifiers, wherein at least one of the sources is controllable by human input; and

means for reconfiguring the complementary classifiers in the distributed face detection system according to the selected classifier parameters in order to detect faces using the distributed face detection system.

23. The system as claimed in claim 22 wherein the complementary classifiers include:

one or more classifiers having high execution speed and featuring operation in one portion of the frequency spectrum of the digital image; and

one or more classifiers having high accuracy and featuring operation in another portion of the frequency spectrum of the digital image.

24. The system as claimed in claim 22 wherein the one or more classifiers having high execution speed and featuring operation in one portion of the frequency spectrum of the digital image comprise a grid pattern classifier.

25. The system as claimed in claim 22 wherein the one or more classifiers having high accuracy and featuring operation in another portion of the frequency spectrum of the digital image comprises at least one of a cascaded classifier and a Bayesian classifier.