Two Stage Detection for Photographic Eye Artifacts

• US 20110115949A1
• Filed: 12/04/2010
• Published: 05/19/2011
• Est. Priority Date: 11/18/2005
• Status: Active Grant

• Current Assignee: FotoNation Limited (Adeia Inc.)
• Original Assignee: Tessera Technologies Ireland Limited (Adeia Inc.)
• Inventors: Steinberg, Eran, Ciuc, Mihai, Corcoran, Peter, Pososin, Alexei
• Granted Patent: US 8,160,308 B2
• Time in Patent Office: Days
• US Class Current: 348/241
• CPC Class Codes: G06T 2207/30216 Redeye defect
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction

Abstract

An image acquisition device includes a first speed-optimized filter for producing a first set of candidate red-eye regions for an acquired image; and a second analysis-optimized filter for operating on the first set of candidate red eye regions and the acquired image.

Claims

1. (canceled)

2. A digital image acquisition device, comprising:

an imaging optic and detector for acquiring an image;

a first speed-optimized red eye filter for producing a first set of candidate red-eye regions for the acquired image;

an image encoder for encoding said acquired image

a display; and

a first image correction module for correcting one or more of said first set of candidate red eye regions in the acquired image to produce a first speed-optimized corrected acquired image;

a second analysis-optimized filter configured to produce a second set of candidate red-eye regions for the acquired image, wherein the second analysis-optimized filter is configured to operate on regions of said array identified as corresponding to regions of said first speed-optimized set of candidate red eye regions, and wherein the second analysis optimized filter is configured to provide a falsing analysis of said regions;

a second image correction module for correcting a second set of candidate red eye regions in the acquired image, different than said first set, to produce a second analysis-optimized corrected acquired image; and

an image store memory for storing therein one or more images each including a header and a main image body, said header comprising said acquired image or said first speed-optimized corrected acquired image, or both, and said main image body comprising said acquired image, said first speed-optimized corrected acquired image, or said second analysis-optimized corrected acquired image.

3. A device according to claim 2, wherein said first speed-optimized filter is configured to operate on a low resolution version of said acquired image.

4. A device according to claim 3, wherein said first speed-optimized filter is configured to operate on a sub-sampled version of said acquired image.

5. A device according to claim 3, wherein said image acquisition device is configured to perform a Discrete Cosine Transformation (DCT) of an acquired image of N*M pixels to produce an array of X*Y DCT blocks, where X and Y are less than N and M, respectively, and wherein said first speed-optimized filter is configured to operate on DCT characteristics of said DCT blocks.

6. A device according to claim 5, in which said filter is configured to selectively operate on the chroma coefficients of said DCT characteristics.

7. A device according to claim 5, in which said filter is configured to operate on the DC coefficients of said DCT characteristics.

8. A device according to claim 5, wherein said image compressor is configured to produce a lossy encoded image from said DCT blocks.

9. A device according to claim 8, wherein said encoded image comprises a JPEG image.

10. A device according to claim 9, wherein said image compressor is configured to store said first set of candidate red-eye regions in a header or a footer, or both, of said JPEG image.

11. A device according to claim 5, wherein said first set of candidate red-eye regions comprises a set of DCT block identifiers.

12. A device according to claim 2, in which said first image correction module is configured to correct said acquired image prior to encoding by said image compressor, and wherein an indication of a region corrected by said first image correction module is stored in association with said corrected image.

13. A device according to claim 2, operable for communication between the device and the second analysis-optimised filter to produce a second set of candidate red-eye regions for said encoded image based on an analysis of said encoded image and said first set of candidate red-eye regions.

14. A device according to claim 2, wherein said device is configured to partially decompress said stored image to produce an array of regions, each corresponding to a one or more pixels of an originally acquired image.

15. A device according to claim 14, wherein said second analysis-optimized filter is configured to fully decompress said identified regions and to analyse said regions at a pixel level.

16. A device according to claim 2, wherein said second analysis-optimized filter is implemented in an image playback module configured to render said compressed image at a relatively high resolution.

17. A device according to claim 16, in which said image playback module is integrally implemented on said digital image acquisition device and wherein said image playback module is configured to display an image on said image display.

18. A device according to claim 16, in which said image playback module is configured to run on a remote device and wherein said digital image acquisition device is configured to communicate said compressed image and said associated first set of candidate red-eye regions to said remote device.

19. A device according to claim 2, wherein said second analysis-optimized filter is optimized for accuracy.

20. A device according to claim 2, wherein the image acquisition device further comprises an image compressor for compressing said acquired image.

21. A method of processing a digital image comprising:

using one or more processors;

filtering an acquired image with a first speed optimized filter to produce a first set of candidate red-eye regions;

encoding said acquired image;

storing said encoded image in association with said first set of candidate red-eye regions for later image processing of said encoded image;

displaying a first corrected image based on said encoding;

correcting one or more of said first set of candidate red eye regions in an acquired image prior to displaying said first corrected image;

activating a second analysis-optimized filter after displaying said first corrected image for producing a second set of candidate red-eye regions for the acquired image; and

operating the second analysis-optimized filter at least on regions of said array identified as corresponding to regions of said first speed-optimized set of candidate red-eye regions, including performing a falsing analysis of said regions; and

storing one or more images in a memory each including a header and a main image body, said header comprising said acquired image or said first speed-optimized corrected acquired image, or both, and said main image body comprising said acquired image, said first speed-optimized corrected acquired image, or said second analysis-optimized corrected acquired image.

22. A method according to claim 21, further comprising operating on a low resolution version of said acquired image.

23. A method according to claim 22, further comprising operating on a sub-sampled version of said acquired image.

24. A method according to claim 21, further comprising:

performing a Discrete Cosine Transformation (DCT) of an acquired image of N*M pixels to produce an array of X*Y DCT blocks, where X and Y are less than N and M, respectively; and

operating on DCT characteristics of said DCT blocks.

25. A method according to claim 24, further comprising selectively operating on the chroma coefficients of said DCT characteristics.

26. A method according to claim 24, further comprising operating on the DC coefficients of said DCT characteristics.

27. A method according to claim 24, further comprising producing a lossy encoded image from said DCT blocks.

28. A method according to claim 27, wherein said encoded image comprises a JPEG image.

29. A method according to claim 28, further comprising storing said first set of candidate red-eye regions in a header or a footer, or both, of said JPEG image.

30. A method according to claim 24, wherein said first set of candidate red-eye regions comprises a set of DCT block identifiers.

31. A method according to claim 21, further comprising:

correcting said acquired image prior to encoding by said image compressor, and

storing an indication of a region corrected by said first image correction module in association with said corrected image.

32. A method according to claim 21, further comprising producing a second set of candidate red-eye regions for said encoded image based on an analysis of said encoded image and said first set of candidate red-eye regions.

33. A method according to claim 21, further comprising partially decompressing said stored image to produce an array of regions, each corresponding to a one or more pixels of an originally acquired image.

34. A method according to claim 33, further comprising fully decompressing said identified regions and analyzing said regions at a pixel level.

35. A method according to claim 21, wherein said second analysis-optimized filter is implemented in an image playback module, and the method further comprises rendering said compressed image at a relatively high resolution.

36. A method according to claim 35, in which said image playback module is integrally implemented on said digital image acquisition device and the method further comprises displaying an image on said image display.

37. A method according to claim 35, further comprising:

running the image playback module on a remote device, and

communicating said compressed image and said associated first set of candidate red-eye regions to said remote device.

38. A method according to claim 21, wherein said second analysis-optimized filter is optimized for accuracy.

39. A method according to claim 21, further comprising compressing said acquired image.

40. One or more processor readable storage devices having processor readable code embodied thereon, said processor readable code for programming one or more processors to perform a method of processing a digital image, the method comprising:

using one or more processors;

filtering an acquired image with a first speed optimized filter to produce a first set of candidate red-eye regions;

encoding said acquired image;

storing said encoded image in association with said first set of candidate red-eye regions for later image processing of said encoded image;

displaying a first corrected image based on said encoding;

correcting one or more of said first set of candidate red eye regions in an acquired image prior to displaying said first corrected image;

activating a second analysis-optimized filter after displaying said first corrected image for producing a second set of candidate red-eye regions for the acquired image; and

operating the second analysis-optimized filter at least on regions of said array identified as corresponding to regions of said first speed-optimized set of candidate red-eye regions, including performing a falsing analysis of said regions; and

storing one or more images in a memory each including a header and a main image body, said header comprising said acquired image or said first speed-optimized corrected acquired image, or both, and said main image body comprising said acquired image, said first speed-optimized corrected acquired image, or said second analysis-optimized corrected acquired image.

41. One or more processor readable storage devices according to claim 40, wherein the method further comprises operating on a low resolution version of said acquired image.

42. One or more processor readable storage devices according to claim 41, wherein the method further comprises operating on a sub-sampled version of said acquired image.

43. One or more processor readable storage devices according to claim 40, wherein the method further comprises:

performing a Discrete Cosine Transformation (DCT) of an acquired image of N*M pixels to produce an array of X*Y DCT blocks, where X and Y are less than N and M, respectively; and

operating on DCT characteristics of said DCT blocks.

44. One or more processor readable storage devices according to claim 43, wherein the method further comprises selectively operating on the chroma coefficients of said DCT characteristics.

45. One or more processor readable storage devices according to claim 43, wherein the method further comprises operating on the DC coefficients of said DCT characteristics.

46. One or more processor readable storage devices according to claim 43, wherein the method further comprises producing a lossy encoded image from said DCT blocks.

47. One or more processor readable storage devices according to claim 46, wherein said encoded image comprises a JPEG image.

48. One or more processor readable storage devices according to claim 47, wherein the method further comprises storing said first set of candidate red-eye regions in a header or a footer, or both, of said JPEG image.

49. One or more processor readable storage devices according to claim 43, wherein said first set of candidate red-eye regions comprises a set of DCT block identifiers.

50. One or more processor readable storage devices according to claim 40, wherein the method further comprises:

correcting said acquired image prior to encoding by said image compressor, and

storing an indication of a region corrected by said first image correction module in association with said corrected image.

51. One or more processor readable storage devices according to claim 40, wherein the method further comprises producing a second set of candidate red-eye regions for said encoded image based on an analysis of said encoded image and said first set of candidate red-eye regions.

52. One or more processor readable storage devices according to claim 40, wherein the method further comprises partially decompressing said stored image to produce an array of regions, each corresponding to a one or more pixels of an originally acquired image.

53. One or more processor readable storage devices according to claim 52, wherein the method further comprises fully decompressing said identified regions and analyzing said regions at a pixel level.

54. One or more processor readable storage devices according to claim 40, wherein said second analysis-optimized filter is implemented in an image playback module, and the method further comprises rendering said compressed image at a relatively high resolution.

55. One or more processor readable storage devices according to claim 54, in which said image playback module is integrally implemented on said digital image acquisition device and the method further comprises displaying an image on said image display.

56. One or more processor readable storage devices according to claim 54, wherein the method further comprises:

running the image playback module on a remote device, and

communicating said compressed image and said associated first set of candidate red-eye regions to said remote device.

57. One or more processor readable storage devices according to claim 40, wherein said second analysis-optimized filter is optimized for accuracy.

58. One or more processor readable storage devices according to claim 40, wherein the method further comprises compressing said acquired image.