Method and apparatus for embedding digital information in digital multimedia data

• US 6,678,389 B1
• Filed: 12/02/1999
• Issued: 01/13/2004
• Est. Priority Date: 12/29/1998
• Status: Expired due to Fees

• Current Assignee: Kent Ridge Digital Labs
• Original Assignee: Kent Ridge Digital Labs
• Inventors: Wu, Jiankang, Sun, Qibin, Deng, Huijie Robert
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Lu, Tom Y.
• Application Number:
  US09/445,142
• Time in Patent Office: 1,503 Days
• Field of Search: 382/100, 375/130, 375/142, 348/475, 235/494
• 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 1/0064 Geometric transfor invariant watermarking, e.g. affine transform invariant
• CPC Class Codes: G06T 2201/0052 Embedding of the watermark in the frequency domain
• CPC Class Codes: H04N 1/32181 using Karhunen-Loeve transforms
• CPC Class Codes: H04N 1/32187 with selective or adaptive application of the additional information, e.g. in selected frequency coefficients
• CPC Class Codes: H04N 1/32192 according to calculated or estimated visibility of the additional information in the image
• CPC Class Codes: H04N 21/23892 involving embedding information at multiplex stream level, e.g. embedding a watermark at packet level
• CPC Class Codes: H04N 21/8358 involving watermark protecting executable software by watermarking G06F21/16; image watermarking in general G06T1/0021; watermarks inserted in still images for transmission purposes H04N1/32144; inserting watermarks during video coding H04N19/467
• CPC Class Codes: H04N 2201/3233 of authentication information, e.g. digital signature, watermark
• CPC Class Codes: H04N 2201/327 which are undetectable to the naked eye, e.g. embedded codes

Abstract

A steganographic method is disclosed to embed an invisible watermark into an image. It can be used for copyright protection, content authentication or content annotation. The technique is mainly based on K-L transform. Firstly a block and cluster step 106 and cluster selection step 108 are performed to enhance the optimization of K-L transform 110 for a given image. Then a watermark is embedded 114 into the selected eigen-clusters. ECC (Error Correction Code) can be employed to reduce the embedded code error rate. The proposed method is characterized by robustness despite the degradation or modification on the watermarked content. Furthermore, the method can be extended to video, audio or other multimedia especially for multimedia databases in which the stored multimedia are categorized by their contents or classes.

Claims

1. A method of embedding digital information in digital multimedia data, wherein said method includes the following steps:

generating one or more transforms from the digital multimedia data or a portion of the digital multimedia data, wherein said transforms are dependent on the digital multimedia data or the portion of the digital multimedia data;

generating transformed digital multimedia data by applying the one or more transforms to the digital multimedia data or the portion of the digital multimedia data;

embedding the digital information in the transformed digital multimedia data; and

generating the digital multimedia data embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded transformed digital multimedia data;

wherein said digital multimedia data is an image and said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_i coefficients of the transformed digital multimedia data, where the number L_i is determined in accordance with HVS analysis.

2. A method as claimed in claim 1, wherein the method further comprises the following steps for selecting said portion of said digital multimedia data:

partitioning the image into blocks;

clustering the partitioned blocks into classes according to a homogeneity criteria; and

selecting one or more of the clusters as said portion of said digital multimedia data in accordance with HVS analysis.

3. A method as claimed in claim 2, wherein the selecting step is determined in accordance with HVS analysis and a user'"'"'s selection.

4. A method as claimed in claim 1, wherein said digital information is a watermark.

5. A method as claimed in claim 1, wherein said method further includes the following step:

storing the digital multimedia data as a reference file together with associated information.

6. A method as claimed in claim 1, wherein said method further includes the following steps

extracting invariant representations of the image;

storing said extracted representations together with associated information in a reference file.

7. A method as claimed in claim 1, wherein said method further includes the following steps

extracting key features from said image;

and

storing the key features together with associated information.

8. A method as claimed in claim 7, wherein said key features include:

the positions of the key features, invariant attributes of the key features, maximum and minimum colour values of the image, the size of the image.

9. A method as claimed in claim 1, wherein said digital information is a copyright notice.

10. A method as claimed in claim 1, wherein said digital information is an annotation of said digital multimedia data.

11. A method as claimed in claim 1, wherein said digital information is authentication information verifying said digital multimedia data.

12. A method of embedding digital information in digital multimedia data, wherein said method includes the following steps:

generating one or more transforms from the digital multimedia data or a portion of the digital multimedia data, wherein said transforms are dependent on the digital multimedia data or the portion of the digital multimedia data;

generating transformed digital multimedia data by applying the one or more transforms to the digital multimedia data or the portion of the digital multimedia data;

embedding the digital information in the one or more transforms or their inverses; and

generating the digital multimedia data embedded with the digital information by applying one or more embedded inverse transforms to the transformed digital multimedia data.

13. A method as claimed in claim 12, wherein said digital multimedia data is an image.

14. A method as claimed in claim 12, wherein the method further comprises the following steps for selecting said portion of said digital multimedia data:

partitioning the image into blocks;

clustering the partitioned blocks into classes according to a homogeneity criteria; and

selecting one or more of the clusters as said portion of said digital multimedia data in accordance with HVS analysis.

15. A method as claimed in claim 14, wherein the selecting step is determined in accordance with HVS analysis and a user'"'"'s selection.

16. A method as claimed in claim 12, wherein said transform is a Karhunen-Loeve transform.

17. A method as claimed in claim 12, wherein said digital multimedia data is an image and said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_ieigenvectors of the Karhunen-Loeve transforms, where the number L_i is determined in accordance with HVS analysis.

18. A method as claimed in claim 12, wherein said digital information is a watermark.

19. A method as claimed in claim 12, wherein said method further includes the following step:

storing the digital multimedia data as a reference file together with associated information.

20. A method as claimed in claim 13, wherein said method further includes the following steps

extracting invariant representations of the image;

storing said extracted representations together with associated information in a reference file.

21. A method as claimed in claim 13, wherein said method further includes the following steps

extracting key features from said image;

and

storing the key features together with associated information.

22. A method as claimed in claim 21, wherein said key features include:

the positions of the key features, invariant attributes of the key features, colour values of the key features, maximum and minimum colour values of the image, the size of the image.

23. A method as claimed in claim 12, wherein said digital information is a copyright notice.

24. A method as claimed in claim 12, wherein said digital information is an annotation of said digital multimedia data.

25. A method as claimed in claim 12, wherein said digital information is authentication information verifying said digital multimedia data.

68. A method as claimed in claim 12, further comprising:

extracting digital information from first digital multimedia data having embedded therein said digital information including the following steps;

deriving one or more embedded transforms from a reference file, wherein said embedded transforms are dependent on second digital multimedia data or a portion thereof;

deriving transformed digital multimedia data from the reference file, wherein said transformed digital multimedia data is dependent on the second digital multimedia data or a portion thereof;

generating transformed digital multimedia data by applying one or more of the derived embedded transforms to the first digital multimedia data or a portion thereof; and

correlating the generated transformed digital multimedia with the derived transformed digital multimedia to check whether the digital information embedded in the first multimedia data is the same as that derived from the reference file.

69. A method as claimed in claim 12, further comprising extracting digital information from first digital multimedia data having embedded therein said digital information, including the following steps:

deriving one or more transforms from a reference file, wherein said transforms are dependent on second digital multimedia data or a portion thereof;

deriving transformed digital multimedia data from the reference file, wherein said transformed digital multimedia data is dependent on the second digital multimedia data or a portion thereof;

generating embedded transformed digital multimedia data by applying one or more of the derived embedded transforms to the first digital multimedia data or a portion thereof; and

correlating the generated embedded transformed digital multimedia with the derived transformed digital multimedia so as to extract the digital information.

26. A method of embedding digital information in a digital image, wherein said method includes the following steps:

partitioning the image into blocks of pixels;

clustering the partitioned blocks into classes according to a homogeneity criteria;

selecting one or more of the clusters in accordance with HVS analysis;

generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

embedding, for each selected cluster, the digital information in the associated K-L transform; and

applying, for each selected cluster, an inverse transform of the embedded associated K-L transform to the transformed image blocks of the selected cluster.

27. A method as claimed in claim 26, wherein said digital information is embedded in the first L_i eigenvectors of the Karhunen-Loeve transforms, where the number L_i is determined in accordance with HVS analysis.

28. A method as claimed in claim 26, wherein said digital information is a noise-like watermark.

29. A method as claimed in claim 26, wherein the selecting step is determined in accordance with HVS analysis and a user'"'"'s selection.

30. A method of embedding digital information in a digital image, wherein said method includes the following steps:

partitioning the image into blocks of pixels;

clustering the partitioned blocks into classes according to a homogeneity criteria;

selecting one or more of the clusters in accordance with HVS analysis;

generating an associated Karhunen-Loeve (K-L) transform for each selected cluster,

wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

embedding, for each selected cluster, the digital information in the selected transformed image blocks; and

applying, for each selected cluster, an inverse transform of the associated K-L transform to the embedded transform image blocks of the selected cluster.

31. A method as claimed in claim 30, wherein said digital information is embedded in the first L_i coefficients of the transformed image blocks, where the number L_i is determined in accordance with HVS analysis.

32. A method as claimed in claim 30, wherein said digital information is a noise-like watermark.

33. A method as claimed in claim 30, wherein the selecting step is determined in accordance with HVS analysis and a user'"'"'s selection.

34. A method of embedding digital information in a digital image, wherein said method includes the following steps:

partitioning the image into blocks of pixels;

clustering the partitioned blocks into classes according to a homogeneity criteria;

selecting one or more of the clusters in accordance with HVS analysis;

generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

selecting, for each selected cluster, either the associated K-L transform or its corresponding transformed image blocks for embedding;

embedding, for each selected cluster, the digital information in the selected K-L transform or the selected transformed image blocks;

applying, for those selected clusters in which the digital information has been embedded in the associated K-L transform, an inverse transform of the embedded associated K-L transform to the transformed image blocks of the selected cluster; and

applying, for those selected clusters in which the digital information has been embedded in the transformed image blocks, an inverse transform of the associated K-L transform to the embedded transform image blocks of the selected cluster.

35. A method as claimed in claim 34, wherein said digital information is embedded in the first L_i eigenvectors of the Karhunen-Loeve transforms, where the number L_i is determined in accordance with HVS analysis.

36. A method as claimed in claim 34, wherein said digital information is embedded in the first L_i coefficients of the transformed image blocks, where the number L_i is determined in accordance with HVS analysis.

37. A method as claimed in claim 34, wherein said digital information is a noise-like watermark.

38. A method as claimed in claim 34, wherein the selecting step is determined in accordance with HVS analysis and a user'"'"'s selection.

62. A method as claimed in claim 34, further comprising registering the image including the following steps:

extracting key points from the image and their invariant attributes;

extracting other global information from the image;

aligning the extracted key points and their invariant attributes with the global information; and

encrypting the aligned information to provide an image signature.

39. A method of embedding digital information in a plurality of images, wherein the images are stored and classified into one or more classes and said method includes the following steps:

generating one or more transforms from the images of said respective one or more classes, wherein said transforms are dependent on the images;

generating transformed images by applying the one or more transforms to the images of said respective one or more classes;

embedding the digital information in the transformed images; and

generating the images data embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded transformed images;

wherein said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_i coefficients of the transformed digital multimedia data, where the number L_i is determined in accordance with HVS analysis.

65. A method as claimed in claim 39, further comprising modifying a first image in accordance with an image signature, wherein the image signature includes encrypted key points of a second image and their invariant attributes and other aligned global information of the second image, wherein said modifying the first image includes the steps of:

decrypting the image signature to obtain the key points and their invariant attributes and other global information of the second image;

extracting key points and their invariant attributes from the first image;

extracting other global information from the first image;

matching the key points and their invariant attributes and other global information of the first and second images; and

normalising the first image based on said matched key points and their invariant attributes and other global information.

40. A method of embedding digital information in a plurality of images, wherein the images are stored and classified into one or more classes and said method includes the following steps:

generating one or more transforms from the images of said respective one or more classes, wherein said transforms are dependent on the images;

generating transformed images by applying the one or more transforms to the images of said respective one or more classes;

embedding the digital information in the one or more transforms or their inverses; and

generating the images embedded with the digital information by applying one or more embedded inverse transforms to the transformed images.

41. A method of embedding digital information into video images including a plurality of frames, wherein said method includes the following steps:

extracting one or more key frames from the video images;

generating one or more transforms from the key frames or a part thereof, wherein said transforms are dependent on the key frames or the part thereof;

generating transformed key frames or part thereof by applying the one or more transforms to the key frames or the part thereof;

embedding the digital information in the transformed key frames or part thereof; and

generating the key frames or part thereof embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded key frames or part thereof;

wherein said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_i coefficients of the transformed digital multimedia data, where the number L_i is determined in accordance with HVS analysis.

42. A method of embedding digital information into video images including a plurality of frames, wherein said method includes the following steps:

extracting one or more key frames from the video images;

generating one or more transforms from the key frames or a part thereof, wherein said transforms are dependent on the key frames or the part thereof;

generating transformed key frames or part thereof by applying the one or more transforms to the key frames or the part thereof;

embedding the digital information in the one or more transforms or their inverses; and

generating the key frames or part thereof embedded with the digital information by applying one or more embedded inverse transforms to the transformed key frames or part thereof.

43. A method of modifying a first image in accordance with an image signature, wherein the image signature includes encrypted key points of a second image and their invariant attributes and other aligned global information of the second image and the method includes the steps of:

decrypting the image signature to obtain the key points and their invariant attributes and other global information of the second image;

extracting key points and their invariant attributes from the first image;

extracting other global information from the first image;

matching the key points and their invariant attributes and other global information of the first and second images; and

normalising the first image based on said matched key points and their invariant attributes and other global information.

44. An apparatus for embedding digital information in digital multimedia data, wherein said apparatus includes:

means for generating one or more transforms from the digital multimedia data or a portion of the digital multimedia data, wherein said transforms are dependent on the digital multimedia data or the portion of the digital multimedia data;

means for generating transformed digital multimedia data by applying the one or more transforms to the digital multimedia data or the portion of the digital multimedia data;

means for embedding the digital information in the transformed digital multimedia data; and

means for generating the digital multimedia data embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded transformed digital multimedia data

wherein said digital multimedia data is an image and said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_i coefficients of the transformed digital multimedia data, where the number L_i is determined in accordance with HVS analysis.

45. An apparatus for embedding digital information in digital multimedia data, wherein said apparatus includes:

means for generating one or more transforms from the digital multimedia data or a portion of the digital multimedia data, wherein said transforms are dependent on the digital multimedia data or the portion of the digital multimedia data;

means for generating transformed digital multimedia data by applying the one or more transforms to the digital multimedia data or the portion of the digital multimedia data;

means for embedding the digital information in the one or more transforms or their inverses; and

means for generating the digital multimedia data embedded with the digital information by applying one or more embedded inverse transforms to the transformed digital multimedia data.

46. An apparatus for embedding digital information in a digital image, wherein said apparatus includes:

means for partitioning the image into blocks of pixels;

means for clustering the partitioned blocks into classes according to a homogeneity criteria;

means for selecting one or more of the clusters in accordance with HVS analysis;

means for generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

means for applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks.

47. An apparatus for embedding digital information in a digital image, wherein said apparatus includes:

means for partitioning the image into blocks of pixels;

means for clustering the partitioned blocks into classes according to a homogeneity criteria;

means for selecting one or more of the clusters in accordance with HVS analysis;

means for generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

means for applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

means for embedding, for each selected cluster, the digital information in the selected transformed image blocks; and

means for applying, for each selected cluster, an inverse transform of the associated K-L transform to the embedded transform image blocks of the selected cluster.

48. An apparatus for embedding digital information in a digital image, wherein said apparatus includes:

means for partitioning the image into blocks of pixels;

means for clustering the partitioned blocks into classes according to a homogeneity criteria;

means for selecting one or more of the clusters in accordance with HVS analysis;

means for generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

means for applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

means for selecting, for each selected cluster, either the associated K-L transform or its corresponding transformed image blocks for embedding;

means for embedding, for each selected cluster, the digital information in the selected K-L transform or the selected transformed image blocks;

means for applying, for those selected clusters in which the digital information has been embedded in the associated K-L transform, an inverse transform of the embedded associated K-L transform to the transformed image blocks of the selected cluster; and

means for applying, for those selected clusters in which the digital information has been embedded in the transformed image blocks, an inverse transform of the associated K-L transform to the embedded transform image blocks of the selected cluster.

63. An apparatus as claimed in claim 48, further comprising means for registering the image including:

means for extricating key points from the image and their invariant attributes;

means for extracting other global information from the image;

means for aligning the extracted key points and their invariant attributes with the global information;

means for encrypting the aligned information to provide an image signature.

49. An apparatus for embedding digital information in a plurality of images, wherein the images are stored and classified into one or more classes and said apparatus includes:

means for generating one or more transforms from the images of said respective one or more classes, wherein said transforms are dependent on the images;

means for generating transformed images by applying the one or more transforms to the images of said respective one or more classes;

means for embedding the digital information in the one or more transforms or their inverses; and

means for generating the images embedded with the digital information by applying one or more embedded inverse transforms to the transformed images.

66. An apparatus as claimed in claim 49, further comprising means for modifying a first image in accordance with an image signature, wherein the image signature includes encrypted key points of a second image and their invariant attributes and other aligned global information of the second image, said means for modifying including:

means for decrypting the image signature to obtain the key points and their invariant attributes and other global information of the second image;

means for extracting key points and their invariant attributes from the first image;

means for extracting other global information from the first image;

means for matching the key points and their invariant attributes and other global information from the first image; and

means for normalising the first image based on said matched key points and their invariant attributes and other global information.

70. An apparatus as claimed in claim 49, further comprising means for extracting digital information from first digital multimedia data having embedded therein said digital information, including:

means for deriving one or more embedded transforms from a reference file, wherein said embedded transforms are dependent on second digital multimedia data or a portion thereof;

means for deriving transformed digital multimedia data from the reference file, wherein said transformed digital multimedia data is dependent on the second digital multimedia data or a portion thereof;

means for generating transformed digital multimedia data by applying one or more of the derived embedded transforms to the first digital multimedia data or a portion thereof; and

means for correlating the generated transformed digital multimedia with the derived transformed digital multimedia to check whether the digital information embedded in the first multimedia data is the same as that derived from the reference file.

50. An apparatus for embedding digital information into video images including a plurality of frames, wherein said apparatus includes:

means for extracting one or more key frames from the video images;

means for generating one or more transforms from the key frames or a part thereof, wherein said transforms are dependent on the key frames or the part thereof;

means for generating transformed key frames or part thereof by applying the one or more transforms to the key frames or the part thereof;

means for embedding the digital information in the transformed key frames or part thereof; and

means for generating the key frames or part thereof embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded key frames or part thereof;

wherein said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_i coefficients of the transformed digital multimedia data, where the number L_i is determined in accordance with HVS analysis.

51. An apparatus for embedding digital information into video images including a plurality of frames, wherein said apparatus includes:

means for extracting one or more key frames from the video images;

means for generating one or more transforms from the key frames or a part thereof, wherein said transforms are dependent on the key frames or the part thereof;

means for generating transformed key frames or part thereof by applying the one or more transforms to the key frames or the part thereof;

means for embedding the digital information in the one or more transforms or their inverses; and

means for generating the key frames or part thereof embedded with the digital information by applying one or more embedded inverse transforms to the transformed key frames or part thereof.

52. An apparatus for modifying a first image in accordance with an image signature, wherein the image signature includes encrypted key points of a second image and their invariant attributes and other aligned global information of the second image and the apparatus includes:

means for decrypting the image signature to obtain the key points and their invariant attributes and other global information of the second image;

means for extracting key points and their invariant attributes from the first image;

means for extracting other global information from the first image;

means for matching the key points and their invariant attributes and other global information of the first and second images; and

means for normalising the first image based on said matched key points and their invariant attributes and other global information.

53. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information in digital multimedia data, wherein said computer program product includes:

means for generating one or more transforms from the digital multimedia data or a portion of the digital multimedia data, wherein said transforms are dependent on the digital multimedia data or the portion of the digital multimedia data;

means for generating transformed digital multimedia data by applying the one or more transforms to the digital multimedia data or the portion of the digital multimedia data;

means for embedding the digital information in the transformed digital multimedia data; and

means for generating the digital multimedia data embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded transformed digital multimedia data;

wherein said digital multimedia data is an image and said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_i coefficients of the transformed digital multimedia data, where the number L_i is determined in accordance with HVS analysis.

54. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information in digital multimedia data, wherein said computer program product includes:

means for generating one or more transforms from the digital multimedia data or a portion of the digital multimedia data, wherein said transforms are dependent on the digital multimedia data or the portion of the digital multimedia data;

means for generating transformed digital multimedia data by applying the one or more transforms to the digital multimedia data or the portion of the digital multimedia data;

means for embedding the digital information in the one or more transforms or their inverses; and

means for generating the digital multimedia data embedded with the digital information by applying one or more embedded inverse transforms to the transformed digital multimedia data.

55. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information in a digital image, wherein said computer program product includes:

means for partitioning the image into blocks of pixels;

means for clustering the partitioned blocks into classes according to a homogeneity criteria;

means for selecting one or more of the clusters in accordance with HVS analysis;

means for generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

means for applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

means for embedding, for each selected cluster, the digital information in the associated K-L transform; and

means for applying, for each selected cluster, an inverse transform of the embedded associated K-L transform to the transformed image blocks of the selected cluster.

56. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information in a digital image, wherein said computer program product includes:

means for partitioning the image into blocks of pixels;

means for clustering the partitioned blocks into classes according to a homogeneity criteria;

means for selecting one or more of the clusters in accordance with HVS analysis;

means for generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

means for applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

means for embedding, for each selected cluster, the digital information in the selected transformed image blocks; and

means for applying, for each selected cluster, an inverse transform of the associated K-L transform to the embedded transform image blocks of the selected cluster.

64. A computer program product as claimed in claim 56, in which the computer program registers the image and includes:

means for extracting key points from the image and their invariant attributes;

means for extracting other global information from the image;

means for aligning the extracted key points and their invariant attributes with the global information;

means for encrypting the aligned information to provide an image signature.

57. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information in a digital image, wherein said computer program product includes:

means for partitioning the image into blocks of pixels;

means for clustering the partitioned blocks into classes according to a homogeneity criteria;

means for selecting one or more of the clusters in accordance with HVS analysis;

means for generating an associated Karhunen-Loeve (K-L) transform for each selected cluster, wherein the image blocks of each selected cluster are used as a vector population for the purposes of generating the associated K-L transform;

means for applying the associated K-L transform to the image blocks of the selected cluster to produce corresponding transformed image blocks;

means for selecting, for each selected cluster, either the associated K-L transform or its corresponding transformed image blocks for embedding;

means for embedding, for each selected cluster, the digital information in the selected K-L transform or the selected transformed image blocks;

means for applying, for those selected clusters in which the digital information has been embedded in the associated K-L transform, an inverse transform of the embedded associated K-L transform to the transformed image blocks of the selected cluster; and

means for applying, for those selected clusters in which the digital information has been embedded in the transformed image blocks, an inverse transform of the associated K-L transform to the embedded transform image blocks of the selected cluster.

58. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information in a plurality of images, wherein the images are stored and classified into one or more classes and said computer program product includes:

means for generating one or more transforms from the images of said respective one or more classes, wherein said transforms are dependent on the images;

means for generating transformed images by applying the one or more transforms to the images of said respective one or more classes;

means for embedding the digital information in the one or more transforms or their inverses; and

means for generating the images embedded with the digital information by applying one or more embedded inverse transforms to the transformed images.

59. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information into video images including a plurality of frames, wherein said computer program product includes:

means for extracting one or more key frames from the video images;

means for generating one or more transforms from the key frames or a part thereof, wherein said transforms are dependent on the key frames or the part thereof;

means for generating transformed key frames or part thereof by applying the one or more transforms to the key frames or the part thereof;

means for embedding the digital information in the transformed key frames or part thereof; and

means for generating the key frames or part thereof embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded key frames or part thereof;

wherein said one or more transforms are Karhunen-Loeve transforms and said digital information is embedded in the first L_i coefficients of the transformed digital multimedia data, where the number L_i is determined in accordance with HVS analysis.

60. A computer program product including a computer readable medium having recorded thereon a computer program for embedding digital information into video images including a plurality of frames, wherein said computer program product includes:

means for extracting one or more key frames from the video images;

means for generating one or more transforms from the key frames or a part thereof, wherein said transforms are dependent on the key frames or the part thereof;

means for generating transformed key frames or part thereof by applying the one or more transforms to the key frames or the part thereof;

means for embedding the digital information in the one or more transforms or their inverses; and

means for generating the key frames or part thereof embedded with the digital information by applying one or more embedded inverse transforms to the transformed key frames or part thereof.

67. A computer program product as claimed in claim 60, including:

means for modifying the image in accordance with the image signature;

wherein the image signature includes encrypted key points of a second image and their invariant attributes and other aligned global information of the second image, said means for modifying including;

means for decrypting the image signature to obtain the key points and their invariant attributes and other global information of the second image;

means for extracting key points and their invariant attributes from the first image;

means for extracting other global information from the first image;

means for matching the key points and their invariant attributes and other global information of the first image based on said matched key points and their invariant attributes and other global information.

71. A computer program product as claimed in claim 60, including means extracting digital information from first digital multimedia data having embedded therein said digital information, including:

means for deriving one or more embedded transforms from a reference file, wherein said embedded transforms are dependent on second digital multimedia data or a portion thereof;

means for deriving transformed digital multimedia data from the reference file, wherein said transformed digital multimedia data is dependent on the second digital multimedia data or a portion thereof;

means for generating transformed digital multimedia data by applying one or more of the derived embedded transforms to the first digital multimedia data or a portion thereof; and

means for correlating the generated transformed digital multimedia with the derived transformed digital multimedia to check whether the digital information embedded in the first multimedia data is the same as that derived from the reference file, having embedded therein said digital information, wherein said computer program product includes;

means for partitioning the image into blocks of pixels;

means for clustering the partitioned blocks into classes according to a homogeneity criteria;

means for selecting one or more of the clusters in accordance with information obtained from a reference file;

means for retrieving, for each selected cluster, an average block of pixels stored in the reference file;

means for averaging, for each selected cluster, the blocks of pixels of the image; and

means for correlating, for each selected pixel, the averaged blocks of the image with the retrieved blocks to determine whether a watermark exists.

61. A computer program product including a computer readable medium having recorded thereon a computer program for modifying a first image in accordance with an image signature, wherein the image signature includes encrypted key points of a second image and their invariant attributes and other aligned global information of the second image and the computer program product includes:

means for decrypting the image signature to obtain the key points and their invariant attributes and other global information of the second image;

means for extracting key points and their invariant attributes from the first image;

means for extracting other global information from the first image;

means for matching the key points and their invariant attributes and other global information of the first and second images; and

means for normalizing the first image based on said matched key points and their invariant attributes and other global information.

72. A method of embedding digital information in a plurality of images, wherein the images are stored and classified into one or more classes and said method includes the following steps:

generating one or more transforms from the images of said respective one or more classes, wherein said transforms are dependent on the images;

generating transformed images by applying the one or more transforms to the images of said respective one or more classes;

embedding the digital information in the transformed images;

generating the images data embedded with the digital information by applying one or more inverse transforms of the one or more transforms to the embedded transformed images; and

modifying a first image in accordance with an image signature, wherein the image signature includes encrypted key points of a second image and their invariant attributes and other aligned global information of the second image, wherein said modifying the first image includes the steps of;

decrypting the image signature to obtain the key points and their invariant attributes and other global information of the second image;

extracting key points and their invariant attributes from the first image;

extracting other global information from the first image;

matching the key points and their invariant attributes and other global information of the first and second images; and

normalising the first image based on said matched key points and their invariant attributes and other global information.