Digital watermarking and fingerprinting including synchronization, layering, version control, and compressed embedding

• US 20030185417A1
• Filed: 01/22/2003
• Published: 10/02/2003
• Est. Priority Date: 01/22/2002
• Status: Active Grant

• Current Assignee: Digimarc Corporation
• Original Assignee: Digimarc Corporation
• Inventors: Levy, Kenneth L., Rhoads, Geoffrey B., Ellingson, Eric E., Alattar, Adnan M., Stager, Reed R.
• Granted Patent: US 7,020,304 B2
• Time in Patent Office: Days
• US Class Current: 382/100
• CPC Class Codes: G06T 1/0064 Geometric transfor invariant watermarking, e.g. affine transform invariant
• CPC Class Codes: G06T 1/0071 using multiple or alternating watermarks
• CPC Class Codes: G06T 1/0085 Time domain based watermarking, e.g. watermarks spread over several images
• CPC Class Codes: G06T 2201/0051 Embedding of the watermark in the spatial domain
• CPC Class Codes: G06T 2201/0052 Embedding of the watermark in the frequency domain
• CPC Class Codes: G06T 2201/0061 Embedding of the watermark in each block of the image, e.g. segmented watermarking
• CPC Class Codes: G06T 2201/0065 Extraction of an embedded watermark; Reliable detection
• CPC Class Codes: G06T 2201/0601 whereby calibration information is embedded in the watermark, e.g. a grid, a scale, a list of transformations

Abstract

This disclosure describes several video watermarking and fingerprinting enhancements. These enhancements include synchronizing watermark detectors with one-dimensional calibration signals, layering digital watermarks, watermarks for version control, compressed domain watermarking, watermarking of video object layers, key channel watermark embedding for video, robust fingerprinting of video and watermarking of scalable video.

Claims

1. A method of detecting a digital watermark comprising:

transforming video data into a one dimensional video signal; and

performing calibration of the one dimensional signal with a one dimensional calibration signal to compensate for geometric distortion of the video signal.

2. The method of claim 1 wherein the transforming includes combining rows within a video frame into a first one dimensional signal.

3. The method of claim 2 wherein the transforming includes combining columns within a video frame into a second one dimensional signal;

and performing calibration of the first and second one dimensional signal with corresponding one dimensional calibration signals.

4. The method of claim 1 wherein the transforming includes averaging rows or columns of one or more video frames.

5. The method of claim 1 wherein performing calibration includes correlating the one dimensional signal with the one dimensional calibration signal.

6. The method of claim 5 wherein correlating includes correlating in a transform domain.

7. The method of claim 6 wherein the calibration signal comprises impulse functions in the transform domain.

8. The method of claim 1 wherein the transforming includes combining rows within a video frame into a first one dimensional signal, and combining columns within a video frame into a second one dimensional signal;

and performing calibration of the first and second one dimensional signal with corresponding one dimensional calibration signals, the calibration being performed iteratively to compute horizontal and vertical scaling and rotation.

9. The method of claim 1 including computing a Fourier transform of rows of video and combining the rows, computing a Fourier transform of columns of video and combining the columns, and searching for horizontal and vertical scaling and rotation of the video in the combined rows and columns using digital watermark calibration patterns.

10. A tangible medium on which is stored instructions for performing the method of claim 1.

11. A method of digitally watermarking a media signal with watermark layers at different times, the method comprising:

receiving a host media signal;

analyzing the host signal for presence of an embedded digital watermark;

based on analyzing the host signal, identifying embedded portions of the host media signal that carry at least a first layer of an embedded digital watermark and available portions that do not carry an embedded digital watermark; and

embedding a second or subsequent layer of an embedded digital watermark in the available portions.

12. The method of claim 11 wherein the first layer includes a calibration digital watermark, and the available portions are identified relative to reference locations provided by the calibration signal watermark.

13. The method of claim 11 wherein the analyzing includes:

sliding a detection window over the host media signal; and

performing digital watermark detection repeatedly in the sliding detection window to measure detection statistics at locations throughout the host signal;

wherein the identifying of available portions includes identifying locations where the detection statistics indicate absence of an embedded digital watermark.

14. The method of claim 13 wherein the detection statistics comprise correlation statistics, and correlation peaks in the correlation statistics provide a reference for identifying available portions in the host signal.

15. The method of claim 13 wherein the detection statistics comprise a measure of message symbol accuracy in an embedded digital watermark.

16. The method of claim 13 wherein the digital watermark detection includes performing detection using two or more pseudo-random keys, each key corresponding to a digital watermark layer.

17. The method of claim 11 wherein the second or subsequent layer carries an embedded version identifier.

18. The method of claim 17 wherein the version identifier comprises a time-date stamp embedded in a digital watermark.

19. The method of claim 17 wherein the version identifier comprises a pseudo-random number that indexes version information in a database.

20. The method of claim 17 wherein the version identifier comprises a sequential number selected from a number sequence, wherein subsequent versions of the host signal are assigned sequential numbers from the number sequence.

21. The method of claim 11 wherein the host signal includes a layer carrying a public watermark;

and another layer carrying a private watermark.

22. The method of claim 21 wherein the public watermark indexes a public database, and the private watermark indexes a private database.

23. The method of claim 21 wherein the public watermark provides copyright communication information, and the private watermark provides content identification information.

24. The method of claim 21 wherein the private watermark conveys a device identifier to trace the host media signal to a particular device.

25. The method of claim 21 wherein the private watermark indexes private information in a database, and the public watermark indexes public information in the database.

26. A tangible medium on which is stored instructions for performing the method of claim 11.

27. A method of version control of media signals using digital watermarking, the method comprising:

embedding a first watermark layer in a media signal conveying a content identifier; and

as new versions of the media signal are created, embedding a second or subsequent watermark layer including a version identifier to distinguish between versions of the media signal having the same content identifier.

28. The method of claim 27 wherein the version identifier comprises a time-date stamp embedded in a digital watermark.

29. The method of claim 27 wherein the version identifier comprises a sequential number selected from a number sequence, wherein subsequent versions of the host signal are assigned sequential numbers from the number sequence.

30. A method of detecting content flags embedded in a host media signal comprising:

receiving the host media signal;

detecting a calibration watermark signal in the host media signal; and

determining a content flag associated with the calibration signal.

31. The method of claim 30 wherein the calibration signal comprises an array of impulses in a transform domain, and different arrays correspond to different content type flags.

32. The method of claim 30 wherein the calibration watermark is used to determine temporal or geometric distortion of the host media signal.

33. The method of claim 30 wherein the calibration signal comprises an array of impulses in a transform domain, and different phase specifications correspond to different content type flags.

34. The method of claim 30 wherein the content flag comprises a content rating for the host signal.

35. The method of claim 30 wherein the content flag includes a control signal for controlling playback, copying, pausing, or redistribution of the host media signal.

36. The method of claim 30 wherein the calibration signal is used to compensate for geometric or temporal distortion to align the host signal and enable a variable message payload watermark to be extracted.

37. The method of claim 30 wherein the calibration signal is used to align the host signal for extraction of a separate watermark layer carrying a variable message payload.

38. The method of claim 30 wherein the calibration signal is formed by patterning of blocks of a digital watermark signal.

39. A tangible medium on which is stored instructions for performing the method of claim 30.

40. A method for video watermarking comprising:

receiving video comprised of two or more video object layers;

separately embedding digital watermarks in each video layer, including embedding a digital watermark with synchronization attributes used to synchronize each of the video layers; and

forming a compressed bit stream comprised of bit streams for each of the watermarked video layers.

41. The method of claim 40 including embedding a master synchronization signal in each of the layers.

42. The method of claim 41 including embedding variable data carrying watermarks in each of the layers.

43. The method of claim 40 including embedding a master synchronization signal in one of the layers such that the master synchronization signal enables synchronization of a digital watermark reader for the digital watermarks in each of the layers.

44. The method of claim 40 wherein at least one of the video object layers comprises a sprite.

45. The method of claim 40 including embedding a different synchronization signal in each of the video object layers.

46. The method of claim 40 including:

parsing a compressed video stream to extract compressed bit streams representing the video object layers;

modifying transform domain coefficients in the compressed bit stream to embed the digital watermarks in the video object layers.

47. A method for detecting a digital watermark in a video signal comprised of video object layers, the method comprising:

detecting a master synchronization signal in the video signal;

using the master synchronization to calibrate detection of watermarks embedded in two or more different video object layers.

48. The method of claim 47 wherein the master synchronization signal is embedded in each of the video object layers.

49. The method of claim 47 wherein the master synchronization signal is embedded in one of the video object layers and is used to synchronize watermark detection in one or more other video object layers.

50. A tangible medium on which is stored instructions for performing the method of claim 47.

51. A method of embedding auxiliary data in a compressed data stream comprising:

providing an auxiliary data stream to be embedded in the compressed data stream; and

selecting a sequence of prediction mode parameters to represent data in the compressed data stream such that the prediction mode parameter sequence corresponds to auxiliary symbols of the auxiliary data stream.

52. The method of claim 51 wherein the compressed data stream corresponds to a compressed video stream and the prediction mode corresponds to prediction for B-type prediction data in the compressed video stream.

53. The method of claim 51 wherein the prediction modes include forward and backward prediction.

54. A tangible medium on which is stored instructions for performing the method of claim 51.

55. A method of extracting an auxiliary data stream from a compressed data stream comprising:

reading parameters identifying prediction mode of data in the compressed data stream; and

extracting the auxiliary data by translating the parameters into auxiliary data symbols that correspond to the parameters.

56. The method of claim 55 wherein the compressed data stream corresponds to a compressed video stream and the prediction mode corresponds to prediction for B-type prediction data in the compressed video stream.

57. The method of claim 55 wherein the prediction modes include forward and backward prediction.

58. A tangible medium on which is stored instructions for performing the method of claim 55.

59. A method for video watermarking in scalable video comprising:

generating a digital watermark signal;

mapping the digital watermark signal to sample locations at different resolutions in the scalable video; and

repeatedly embedding a digital watermark in the video at the different resolutions.

60. The method of claim 59 wherein the different resolutions are different spatial resolutions.

61. The method of claim 59 wherein the different resolutions are different temporal resolutions.

62. The method of claim 59 wherein the different resolutions are different temporal and spatial resolutions.

63. A tangible medium on which is stored instructions for performing the method of claim 59.

64. A method of video fingerprinting comprising:

computing statistics for video frames;

forming a fingerprint of the video as a change in the statistics over the video frames.

65. The method of claim 64 wherein the statistics comprise averages.

66. The method of claim 65 wherein the averages comprise luminance averages.

67. The method of claim 64 wherein the statistics comprise variances.

68. The method of claim 64 wherein a calibration signal is used to align the video before computing the fingerprint.

69. The method of claim 68 wherein the calibration signal comprises an embedded digital watermark in the video.

70. The method of claim 64 including using the fingerprint to look up related information in a database.

71. A tangible medium having instructions for performing the method of claim 64.

72. A method for detecting a digital watermark in a compressed video signal comprising:

parsing the compressed video signal to extract DCT coefficients; and

performing a combined inverse DCT and watermark demodulation operation on the DCT coefficients.

73. A tangible medium on which is stored instructions for performing the method of claim 72.

74. The method of claim 72 wherein the combined inverse DCT and watermark demodulation operation produces estimates of an error correction coded message;

and

performing error correction decoding on the estimates.

75. A method of embedding a digital watermark in a video stream comprising:

generating a digital watermark signal;

storing the digital watermark signal in a key channel; and

superimposing the digital watermark on video data through a key channel combiner.

76. The method of claim 75 including perceptually adapting the digital watermark based on the video signal.

77. The method of claim 75 including modifying the digital watermark signal so that the digital watermark is transformed into a key channel signal for combining with a video signal using key channel combiner circuitry.

78. A tangible medium on which is stored instructions for performing the method of claim 75.