Real-time face tracking with reference images

• US 8,509,496 B2
• Filed: 11/16/2009
• Issued: 08/13/2013
• Est. Priority Date: 08/11/2006
• Status: Active Grant

• Current Assignee: FotoNation Limited (Adeia Inc.)
• Original Assignee: DigitalOptics Corporation Europe Limited (Adeia Inc.)
• Inventors: Steinberg, Eran, Corcoran, Peter, Pososin, Alexei, Drimbarean, Alexandru, Nanu, Florin, Petrescu, Stefan, Bigiol, Petronel
• Primary Examiner(s): Alavi, Amir
• Application Number:
  US12/619,587
• Publication Number:
  US 20100060727A1
• Time in Patent Office: 1,366 Days
• Field of Search: 382/103, 382/115, 382117-118, 382/164, 382/173, 382/181, 382/190, 382/218, 382/224, 382/254, 382276-277, 382/284, 382/298, 382/299, 382/305, 358/538
• US Class Current: 382/118
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 7/251 involving models
• CPC Class Codes: G06V 10/62 relating to a temporal dimension, e.g. time-based feature extraction; Pattern tracking
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation
• CPC Class Codes: G06V 40/166 using acquisition arrangements
• CPC Class Codes: G06V 40/167 using comparisons between temporally consecutive images

Abstract

A method of tracking a face in a reference image stream using a digital image acquisition device includes acquiring a full resolution main image and an image stream of relatively low resolution reference images each including one or more face regions. One or more face regions are identified within two or more of the reference images. A relative movement is determined between the two or more reference images. A size and location are determined of the one or more face regions within each of the two or more reference images. Concentrated face detection is applied to at least a portion of the full resolution main image in a predicted location for candidate face regions having a predicted size as a function of the determined relative movement and the size and location of the one or more face regions within the reference images, to provide a set of candidate face regions for the main image.

Claims

1. A method of tracking a face in a reference image stream, comprising:

using a digital image acquisition device including an imaging optical system, and a processor, for obtaining meta-data to apply to a further action, wherein the method includes;

acquiring a full resolution main image and an image stream of relatively low resolution reference images each including one or more face regions;

identifying one or more face regions within two or more of said reference images;

determining a relative movement between said two or more reference images;

determining a size and location of said one or more face regions within each of said two or more reference images;

applying concentrated face detection to at least a portion of said full resolution main image in a predicted location for candidate face regions having a predicted size as a function of the determined relative movement and the size and location of said one or more face regions within said reference images, to provide a set of candidate face regions for said main image; and

applying meta-data of one or more of the set of candidate face regions to a further action.

2. A method as claimed in claim 1 wherein said indication of relative movement comprises an amount and direction of movement.

3. A method as claimed in claim 1 wherein said adjusted face detection comprises:

prior to applying face detection, shifting said associated set of candidate face regions as a function of said relative movement.

4. A method as claimed in claim 3, further comprising shifting said face regions as a function of their size and as a function of said relative movement.

5. A method as claimed in claim 1 wherein said adjusted face detection comprises:

applying face detection to a region of a next acquired image comprising candidate regions corresponding to the previously acquired image expanded as a function of said relative movement.

6. A method as claimed in claim 5, wherein said candidate regions of said next acquired image are expanded as a function of their original size and as a function of said relative movement.

7. A method as claimed in claim 1, further comprising selectively apply face recognition using a database to at least some of said candidate face regions to provide an identifier for each of one or more faces recognized in the candidate face regions;

and storing said identifier for said each recognized face in association with at least one image of said image stream.

8. A method as claimed in claim 1 further comprising tracking candidate face regions of different sizes from a plurality of images of said image stream.

9. A method as claimed in claim 1 further comprising merging said set of candidate face regions with one or more previously detected face regions to provide a set of candidate face regions having different parameters.

10. A method as claimed in claim 1, wherein the method is performed periodically on a selected plurality of images of a reference image stream, wherein said plurality of images include a full resolution main acquired image chronologically following a plurality of preview images.

11. A method as claimed in claim 1, further comprising displaying a main acquired image and superimposing one or more indications of one or more tracked candidate face regions on said displayed acquired image.

12. A method as claimed in claim 11 further comprising storing at least one of the size and location of one or more of said set of candidate face regions in association with said main acquired image.

13. A method a claimed in claim 1, wherein said stream of reference images comprises a stream of preview images.

14. A digital image acquisition device for detecting faces in an image stream including one or more optics and a sensor for acquiring said image stream, and a processor for being programmed by a processor-readable medium having digital code embedded therein to perform a method of tracking faces in an image stream, wherein the method comprises:

receiving a new acquired image from a reference image stream including one or more face regions;

receiving an indication of relative movement of said new acquired image relative to a previously acquired image of said reference image stream, said previously acquired image having an associated set of candidate face regions each having a given size and a respective location; and

applying adjusted face detection to at least a portion of said new acquired image in the vicinity of said candidate face regions as a function of said movement, to provide an updated set of candidate face regions applying meta-data of one or more of the set of candidate face regions to a further action.

15. The image acquisition device as claimed in claim 14, further comprising a motion sensor, said motion sensor comprising an accelerometer and a controlled gain amplifier connected to said accelerometer, said apparatus being arranged to set the gain of said amplifier relatively low for acquiring a high resolution image and to set the gain of said amplifier relatively high during acquisition of a stream of relatively low resolution images.

16. The image acquisition device as claimed in claim 14 including a motion sensor, said motion sensor comprising a MEMS sensor.

17. The image acquisition device as claimed in claim 14, wherein the method further comprises selectively applying face recognition using a database to at least some of said candidate face regions to provide an identifier for a face recognized in a candidate face region, and storing the identifier for the recognized face in association with the new acquired image.

18. The image acquisition device as claimed in claim 14, wherein the method further comprises merging new candidate face regions with one or more previously detected face regions to provide said updated set of candidate face regions.

19. The image acquisition device as claimed in claim 14, wherein the method further comprises displaying an acquired image;

and

superimposing an indication of a tracked candidate face region on the displayed acquired image.

20. The image acquisition device as claimed in claim 19, wherein the method further comprises storing at least one of the size and location of a candidate face region in association with the new acquired image.

21. The image acquisition device of as claimed in claim 14, wherein said stream of reference images comprises a stream of preview images.

22. A method of detecting faces in a reference image stream using a digital image acquisition device comprising:

receiving a first acquired image from said reference image stream including one or more face regions;

sub-sampling said first acquired image at a specified resolution one or more times to provide one or more sub-sampled images;

identifying one or more regions of said first acquired image including said one or more face regions within said one or more subsampled images of said first acquired image with probabilities each above a predetermined threshold;

determining a respective size and location of each identified face region within said first acquired image;

receiving a second acquired image from said reference image stream, and sub-sampling and applying face detection to one or more regions of said subsequent acquired image calculated as probably including one or more face regions corresponding to said one or more face regions identified in said first acquired image;

acquiring a full resolution main image and applying meta-data of at least one of the one or more face regions to a further action.

23. A method as claimed in claim 22 wherein said identifying is performed on said sub-sampled image.

24. A method as claimed in claim 22 wherein said face detection is performed with relaxed face detection parameters.

25. A method as claimed in claim 22 wherein each new acquired image is acquired with progressively increased exposure parameters until at least one candidate face region is detected.

26. A method as claimed in claim 22, further comprising tracking candidate face regions of different parameters from a plurality of images of said image stream.

27. A digital image acquisition device for detecting faces in an image stream including one or more optics and a sensor for acquiring said image stream, a processor, and a processor-readable medium having digital code embedded therein for programming the processor to perform a method of detecting faces in the image stream, wherein the method comprises:

receiving a first acquired image from said reference image stream including one or more face regions;

sub-sampling said first acquired image at a specified resolution one or more times to provide one or more sub-sampled images;

identifying one or more regions of said first acquired image including said one or more face regions within said one or more subsampled images of said first acquired image with probabilities each above a predetermined threshold;

determining a respective size and location of each identified face region within said first acquired image;

receiving a second acquired image from said reference image stream, and sub-sampling and applying face detection to one or more regions of said subsequent acquired image calculated as probably including one or more face regions corresponding to said one or more face regions identified in said first acquired image;

acquiring a full resolution main image and applying meta-data of at least one of the one or more face regions to a further action.

28. A device as claimed in claim 27 wherein said identifying is performed on said sub-sampled image.

29. A device as claimed in claim 27 wherein said face detection is performed with relaxed face detection, parameters.

30. A device as claimed in claim 27 where in a face detection mode of said digital image acquisition device, each new acquired image is acquired with progressively increased exposure parameters until at least one candidate face region is detected.

31. A device as claimed in claim 27, wherein the method further comprises selectively applying face recognition using a database to said candidate face regions to provide an identifier for a face recognized in a candidate face region;

and storing the identifier for the recognized face in association with the main image.

32. A device as claimed in claim 27 wherein the method further comprises tracking one or more candidate face regions within a plurality of images of said image stream.

33. One or more non-transitory processor-readable media having code embedded therein for programming a processor to perform a method of tracking a face in a reference image stream, wherein the method comprises:

programming a processor-based digital image acquisition device including an imaging optical system, and a processor, for obtaining meta-data to apply to a further action, wherein the method further includes;

acquiring a full resolution main image and an image stream of relatively low resolution reference images each including one or more face regions;

identifying one or more face regions within two or more of said reference images;

determining a relative movement between said two or more reference images;

determining a size and location of said one or more face regions within each of said two or more reference images;

applying concentrated face detection to at least a portion of said full resolution main image in a predicted location for candidate face regions having a predicted size as a function of the determined relative movement and the size and location of said one or more face regions within said reference images, to provide a set of candidate face regions for said main image; and

applying meta-data of one or more of the set of candidate face regions to a further action.

34. The one or more processor-readable media as claimed in claim 33 wherein said indication of relative movement comprises an amount and direction of movement.

35. The one or more processor-readable media as claimed in claim 33 wherein said adjusted face detection comprises:

prior to applying face detection, shifting said associated set of candidate face regions as a function of said movement.

36. The one or more processor-readable media as claimed in claim 35, wherein the method further comprises shifting said face regions as a function of their size and as a function of said movement.

37. The one or more processor-readable media as claimed in claim 33 wherein said adjusted face detection comprises:

applying face detection to a region of a next acquired image comprising candidate regions corresponding to the previously acquired image expanded as a function of said movement.

38. The one or more processor-readable media as claimed in claim 37, wherein said candidate regions of said next acquired image are expanded as a function of their original size and as a function of said movement.

39. The one or more processor-readable media as claimed in claim 33, wherein the method further comprises selectively apply face recognition using a database to at least some of said candidate face regions to provide an identifier for each of one or more faces recognized in the candidate face regions;

and

storing said identifier for said each recognized face in association with at least one image of said image stream.

40. The one or more processor-readable media as claimed in claim 33 wherein the method further comprises tracking candidate face regions of different sizes from a plurality of images of said image stream.

41. The one or more processor-readable media as claimed in claim 33 wherein the method further comprises merging said set of candidate face regions with one or more previously detected face regions to provide a set of candidate face regions having different parameters.

42. The one or more processor-readable media as claimed in claim 33, wherein the method is performed periodically on a selected plurality of images of a reference image stream, wherein said plurality of images include a full resolution main acquired image chronologically following a plurality of preview images.

43. The one or more processor-readable media as claimed in claim 33, wherein the method further comprises displaying a main acquired image and superimposing one or more indications of one or more tracked candidate face regions on said displayed acquired image.

44. The one or more processor-readable media as claimed in claim 43 wherein the method further comprises storing at least one of the size and location of one or more of said set of candidate face regions in association with said main acquired image.

45. The one or more processor-readable media a claimed in claim 33, wherein said stream of reference images comprises a stream of preview images.

46. One or more non-transitory processor-readable media having code embedded therein for programming a processor to perform a method of detecting faces in a reference image stream using a digital image acquisition device, wherein the method comprises:

programming a processor-based digital image acquisition device including an imaging optical system, and a processor, for obtaining meta-data to apply to a further action, wherein the method further includes;

receiving a first acquired image from said reference image stream including one or more face regions;

sub-sampling said first acquired image at a specified resolution one or more times to provide one or more sub-sampled images;

identifying one or more regions of said first acquired image including said one or more face regions within said one or more subsampled images of said first acquired image with probabilities each above a predetermined threshold;

determining a respective size and location of each identified face region within said first acquired image;

receiving a second acquired image from said reference image stream, and sub-sampling and applying face detection to one or more regions of said subsequent acquired image calculated as probably including one or more face regions corresponding to said one or more face regions identified in said first acquired image;

acquiring a full resolution main image and applying meta-data of at least one of the one or more face regions to a further action.

47. The one or more processor-readable media as claimed in claim 46 wherein said identifying is performed on said sub-sampled image.

48. The one or more processor-readable media as claimed in claim 46 wherein said face detection is performed with relaxed face detection parameters.

49. The one or more processor-readable media as claimed in claim 46 wherein each new acquired image is acquired with progressively increased exposure parameters until at least one candidate face region is detected.

50. The one or more processor-readable media as claimed in claim 46, further comprising tracking candidate face regions of different parameters from a plurality of images of said image stream.