Apparatus and methods for improving detection of watermarks in content that has undergone a lossy transformation

• US 6,782,116 B1
• Filed: 11/04/2002
• Issued: 08/24/2004
• Est. Priority Date: 11/04/2002
• Status: Expired due to Term

• Current Assignee: Thomson Licensing (Vantiva SA)
• Original Assignee: MediaSec Technologies GmbH (Vantiva SA)
• Inventors: Picard, Justin, Thorwirth, Niels, Zhao, Jian
• Primary Examiner(s): Patel, Jayanti K.
• Application Number:
  US10/287,206
• Time in Patent Office: 659 Days
• Field of Search: 382/100, 382/232, 382/244, 382/112, 382/138, 380/203, 380/205, 380/54, 713/176, 713/179, 358/3.28, 358/1.9, 704/233, 704/270
• US Class Current: 382/100
• CPC Class Codes: G06T 1/0028 Adaptive watermarking, e.g. Human Visual System [HVS]-based watermarking
• CPC Class Codes: G06T 1/005 Robust watermarking, e.g. average attack or collusion attack resistant
• CPC Class Codes: G06T 2201/0065 Extraction of an embedded watermark; Reliable detection
• CPC Class Codes: G06T 2201/0081 whereby both original and watermarked images are required at decoder, e.g. destination-based, non-blind, non-oblivious
• CPC Class Codes: G07D 7/0032 using holograms

Abstract

Techniques for improving detection of watermarks in content that has undergone a lossy transformation. One of the techniques is used when the message that is contained in a watermark belonging to a digital representation that is derived from an original watermarked digital representation cannot be decoded. The technique obtains information about the watermark by comparing the watermark vector for the watermark that cannot be decoded with a replica of the watermark vector from the original watermarked digital representation. The replica is made using the message. Depending on the degree of similarity, the watermark'"'"'s presence and some of its characteristics may be determined. Another technique improves the robustness of watermarks that are used for authentication by employing a short (even single-bit) watermark vector to make the watermark and using the message needed for the authentication to determine where the watermark is located in the digital representation. Authentication of a digital representation is done by determining whether the watermark is present in the digital representation. In another technique, detection of the presence of a watermark is used to determine what areas of a digital representation have been subject to alteration. Techniques for synchronizing digital representations for watermark detection and other purposes include adding marks whose locations can be automatically detected only with the help of information that is external to the digital representation, such as a key, and adding marks to a sequence of digital representations and detecting the marks by summing the sequence.

Claims

1. A method of obtaining information about a first digital representation that may be derived from a second digital representation, the second digital representation having a watermark embedded therein and the method comprising the steps of:

extracting a first watermark vector from a portion of the first digital representation that corresponds to a portion of the second digital representation in which the watermark is embedded;

obtaining a second watermark vector that is a replica of the watermark'"'"'s watermark vector; and

computing a similarity of the first watermark vector with the second watermark vector.

2. The method set forth in claim 1 wherein:

in the step of obtaining the second watermark vector, the second watermark vector is a copy of the first watermark'"'"'s watermark vector.

3. The method set forth in claim 1 wherein

the step of obtaining the second watermark vector comprises the steps of:

obtaining information from which the second watermark vector may be computed; and

computing the second watermark vector from the information.

4. The method set forth in claim 3 wherein:

the information from which the second watermark vector may be computed includes a message that is encoded in the watermark.

5. The method set forth in claim 4 wherein:

the message is obtained from a document from which the first digital representation was obtained.

6. The method set forth in claim 4 wherein:

the message is obtained from a source external to a document from which the first digital representation is obtained.

7. The method set forth in claim 1 further comprising the step of:

using the similarity to determine whether an unreadable version of the watermark is present in the portion of the first digital representation.

8. The method set forth in claim 1 further comprising the step of:

using the similarity to determine how an unreadable version of the watermark that is present in the portion of the first digital representation is related to the watermark.

9. The method set forth in claim 1 further comprising the step of:

using the similarity to locate an area where the first digital representation has been altered.

10. The method set forth in claim 1 further comprising the step of:

using the similarity to determine the watermark'"'"'s strength in an area in the first digital representation.

11. The method set forth in claim 1 wherein:

the first digital representation is made from an analog image that is made from the second digital representation.

12. The method set forth in claim 1 wherein:

the first digital representation is derived by a lossy transformation from the second digital representation.

13. Apparatus for obtaining information about a first digital representation that may be derived from a second digital representation that has a watermark embedded therein, the apparatus being implemented in a system that includes a memory and a processor and comprising:

a watermark reader; and

a watermark comparator,

the watermark reader extracting a first watermark vector from a portion of the first digital representation that corresponds to a portion of the second digital representation in which the watermark is embedded and the comparator computing a similarity of the first watermark vector with a second watermark vector which is a replica of the watermark'"'"'s watermark vector.

14. A method of watermarking a first digital representation, the method comprising the steps of:

adding a watermark to the digital representation, the locations at which the watermark is embedded in the digital representation being determined at least in part by a message; and

making the message available to determine locations in a second digital representation, the second digital representation being possibly derived from the first digital representation and the locations being locations at which the watermark can be expected to be embedded.

15. The method set forth in claim 14 wherein:

the contents of the watermark indicate nothing beyond the watermark'"'"'s presence.

16. The method set forth in claim 14 wherein:

the locations at which the watermark is embedded are determined using a key that is a function at least of the message.

17. The method set forth in claim 16 further comprising the step of:

adding an additional watermark to the digital representation, the additional watermark being made using the message, but by another method.

18. A method of embedding a watermark representing a message in a digital representation, the watermark'"'"'s watermark vector containing at least 1 bit and the method comprising the steps of:

making a key that is a function of at least the message; and

using the key to embed the watermark vector in the digital representation.

19. The method set forth in claim 18 wherein:

the key is a secret key.

20. The method set forth in claim 18 wherein:

the watermark vector contains only 1 bit.

21. A method of determining whether a particular digital representation is derived from a watermarked digital representation whose watermark was embedded at locations determined at least in part by a message, the method comprising the steps of:

obtaining the message;

using the message to determine the locations at which the watermark cam be expected to be embedded in the particular representation; and

determining from values at the determined locations whether the watermark is present in the particular digital representation.

22. The method set forth in claim 21 wherein:

the message is obtained from a document from which the particular digital representation was obtained.

23. A method of determining a portion that has been altered in a particular digital representation that is derived from a digital representation that has been watermarked with a robust watermark, the method comprising the steps of:

determining locations in the particular digital representation that should contain the robust watermark; and

determining the portion that has been altered from a less-than-expected frequency of occurrence of the robust watermark at the locations in the portion.

24. The method set forth in claim 23 wherein:

the watermark'"'"'s content indicates nothing beyond the presence of the watermark.

25. The method set forth in claim 23 wherein:

the watermark'"'"'s watermark vector is short.

26. A method of embedding watermarks in a digital representation and/or reading the embedded watermarks, the method comprising the steps of:

analyzing the digital representation to determine favorable locations for the watermarks;

making a mask for the digital representation that specifies the favorable locations; and

using the mask to embed the watermarks in the digital representation and/or read the embedded watermarks.

27. The method set forth in claim 26 wherein:

the steps of analyzing and making the mask are performed for a given digital representation when the watermark is embedded and/or the embedded watermark is read.

28. The method set forth in claim 26 wherein:

the steps of analyzing and making the mask are performed on a digital representation that belongs to a class thereof; and

the mask is employed when a watermark is embedded in a digital representation belonging to the class is watermarked and/or a watermark that has been embedded in a digital representation belonging to the class is read.

29. The method set forth in claim 26 wherein:

in the step of analyzing, the digital representation is analyzed according to the semantic importance of areas in the digital representation.

30. The method set forth in claim 26 wherein:

in the step of analyzing, the digital representation is analyzed according to how the digital representation is processed after the watermark is added.

31. The method set forth in claim 26 wherein:

in the step of analyzing, the image is analyzed according to an optical characteristic of the digital representation.

32. The method set forth in claim 26 wherein:

the mask is used in watermarking the image; and

in the step of analyzing, the digital representation is analyzed according to a factor that may affect reading of the embedded watermark.

33. An improved digital representation, the improved digital representation being derived from another digital representation and the improved digital representation comprising:

an original content of the improved digital representation that is derived from the other digital representation; and

a synchronization pattern of marks in the improved digital representation, the marks being made by altering the original content and the pattern having the characteristic that location information external to the improved digital representation is necessary to automatically detect the marks in the pattern.

34. The improved digital representation set forth in claim 33 wherein:

the synchronization pattern has substantially more marks than a minimum number of marks required to synchronize the improved digital representation.

35. The improved digital representation set forth in claim 33 wherein:

a location for a mark is selected according to a characteristic of the digital representation at the location.

36. The improved digital representation set forth in claim 33 wherein:

the content of the digital representation at location of a mark in the digital representation is modified to make automatic detection of the mark easier.

37. The improved digital representation set forth in claim 33 wherein:

the information external to the improved representation specifies locations of the marks.

38. The improved digital representation set forth in claim 37 further comprising:

a structure in the digital representation; and

the information external to the improved representation specifies the locations of the marks relative to the structure.

39. The improved digital representation set forth in claim 37 wherein:

the information external to the improved digital representation is a key that specifies the locations of the marks.

40. The improved digital representation set forth in claim 33 wherein:

under normal conditions, the pattern is not perceptible to a viewer of an image made from the digital representation.

41. The improved digital representation set forth in claim 33 wherein:

the improved digital representation further contains a watermark; and

the pattern is used in locating the watermark.

42. A method of deriving a digital representation from another digital representation, the method comprising the steps of:

deriving the content of the digital representation from the other digital representation;

adding a synchronization pattern of marks to the digital representation, the marks being made by altering the content and the pattern having the characteristic that location information external to the improved digital representation is necessary to automatically detect the marks in the pattern.

43. The method set forth in claim 42 wherein:

the synchronization pattern has substantially more marks than a minimum number of marks required to synchronize the improved digital representation.

44. The method set forth in claim 42 wherein:

a location for a mark is selected according to a characteristic of the digital representation at the location.

45. The method set forth in claim 42 wherein:

the content of the digital representation at location of a mark in the digital representation is modified to make automatic detection of the mark easier.

46. The method set forth in claim 42 wherein:

the information external to the improved representation specifies locations of the marks.

47. The method set forth in claim 46 further comprising:

a structure in the digital representation; and

the information external to the improved representation specifies the locations of the marks relative to the structure.

48. The method set forth in claim 46 wherein:

the information external to the improved digital representation is a key that specifies the locations of the marks.

49. The method set forth in claim 42 wherein:

under normal conditions, the pattern is not perceptible to a viewer of an image made from the digital representation.

50. The method set forth in claim 42 wherein:

the improved digital representation further contains a watermark; and

the pattern is used in locating the watermark.

51. A method of adding synchronization marks to a sequence of digital representations comprising the steps performed in each image of:

selecting the same predetermined portions of the content of each digital representation as the synchronization marks; and

changing the value of each selected portion by a predetermined amount, whereby digital representations in the sequence may be summed to detect the synchronization marks.

52. The method set forth in claim 51 wherein:

in the step of changing the value of each selected portion, the predetermined amount may alternate between a positive amount and a negative amount in the digital representations belonging to the sequence.

53. The method set forth in claim 51 wherein:

in the step of selecting the same predetermined ones of the portions, the selected portions are selected according to a secret key.

54. The method set forth in claim 51 further comprising the step of:

modifying portions in the area of the selected portion to make the selected portion easier to detect.

55. A method of reversing a transformation in a digital representation that belongs to a sequence of digital representations that has experienced the transformation, the method comprising the steps of:

summing the content of the sequence of transformed images to detect synchronization marks therein that were present in the untransformed sequence;

making a comparison of the positions of the detected synchronization marks in the transformed sequence with the positions of the corresponding synchronization marks in the untransformed sequence; and

reversing the transformation of the digital representation as indicated by the comparison.

56. A method of calibrating a digital imaging device with regard to an analog document that contains a first mark, the method comprising the steps of:

making a digital representation of the analog document, the digital representation being made such that a second mark corresponding to the first mark is visible in the digital representation;

providing a third mark in the digital imaging device; and

calibrating the digital imaging device by matching the second mark and the third mark.

57. The method set forth in claim 56 wherein:

the first mark is only visible under special illumination; and

in the step of making the digital representation, the capture is done at least in part using the special illumination.

58. The method set forth in claim 56 wherein:

in the step of calibrating, the second mark and the third mark are matched as regards location.

59. The method set forth in claim 56 wherein:

in the step of calibrating, the second mark and the third mark are matched as regards appearance.

60. The method set forth in claim 56 wherein:

in the step of calibrating, the matching is done by a user of the digital imaging device.

61. The method set forth in claim 56 wherein:

in the step of calibrating, the matching is done automatically by the digital imaging device.

62. The method set forth in claim 56 wherein:

in the step of calibrating, a first stage of the calibration is done by a user of the digital imaging device and

a second stage of the calibration is done automatically by the digital imaging device.