System and method for digital watermarking
A digital watermarking system and method are disclosed. In one respect, the disclosed digital watermarking includes generating an extracted signal by applying a watermark extractor to an original image, generating a mixed signal by mixing the first signal with a periodic watermark signal using a local weighting factor for the periodic watermark signal that attenuates a strength of the watermark signal in proportion to a pixel luminance level, and replacing the extracted signal in the original image with the mixed signal to generate a marked image, wherein the watermark signal is extractable from the marked image using the watermark extractor.
1. A digital watermarking method, comprising
generating an extracted signal by applying a watermark extractor to an original image;
generating a mixed signal by mixing the extracted signal with a periodic watermark signal using a local weighting factor for the periodic watermark signal that attenuates a strength of the periodic watermark signal in proportion to a pixel luminance level; and
replacing the extracted signal in the original image with the mixed signal to generate a marked image.
2. The method of claim 1 , comprising generating the local weighting factor using a noise visibility function having a gradient removal factor that prevents the noise visibility function from significantly incrementing at locations of spatial gradient in the original image.
3. The method of claim 1 , comprising generating the local weighting factor using a noise visibility function having a density filter that prevents the periodic watermark signal from becoming visible in the marked image at spatial locations with highly dense textures.
4. The method of claim 1 , wherein mixing the extracted signal with the periodic watermark signal includes applying a watermark block of size M×M to corresponding M×M sections of the extracted signal.
5. The method of claim 4 , wherein the watermark block is generated by operations including
generating a plurality of pseudo random numbers using a unique input identification as a seed;
generating a N(0,1) normal distributed random number in double precision using one of the plurality of pseudo random numbers as a seed;
using the normal distributed random number to populate a watermark pattern sequentially; and
applying low-pass filtering of kernel size 3 on the watermark block to limit the frequency band.
6. The method of claim 5 , wherein
the watermark block is of size 64×64; and
the operation of generating a plurality of pseudo random numbers includes generating 4096 random numbers.
7. The method of claim 5 , wherein the unique input identification is 32 bits in length.
8. The method of claim 1 , wherein the periodic watermark signal is a first periodic watermark signal, the method comprising
generating a third signal by applying the watermark extractor to the marked image;
generating a fourth signal by mixing the third signal with a second periodic watermark signal using a local weighting factor for the second periodic watermark signal that attenuates a strength of the second periodic watermark signal in proportion to a pixel luminance level; and
replacing the third signal in the marked image with the fourth signal to generate an additional marked image.
9. The method of claim 8 , wherein the operation of replacing the extracted signal in the original image with the mixed signal and the operation of replacing the third signal in the marked image with the fourth signal occur at different stages of distribution of the original images.
10. The method of claim 1 , wherein the periodic watermark signal is extractable from the marked image using the watermark extractor.
11. The method of claim 1 , wherein the original image is an individual frame of a digital video.
12. The method of claim 1 , wherein the original image is an individual frame of a streaming video and the method is selectively applied on either I frames or subsequent P frames of a streaming video.
13. The method of claim 1 , wherein the watermark extractor uses a Wiener filter.
14. The method of claim 1 , wherein the periodic watermark signal is normalized.
15. The method of claim 8 , wherein either or both of the first and second periodic watermark signals are extractable from the additional marked images using the watermark extractor.
16. A digital watermarking method, comprising
applying an electro-optical transfer function to an original image to generate an EOTF image;
generating an extracted signal by applying a watermark extractor to the EOTF image;
generating a mixed signal by mixing the extracted signal with a periodic watermark signal using a local weighting factor for the periodic watermark signal that attenuates a strength of the watermark signal in proportion to a pixel luminance level;
replacing the extracted signal in the EOTF image with the mixed signal to generate a marked image; and
applying an optical-electro transfer function to the marked image to generate a final marked image, wherein the optical-electro transfer function utilized performs an inverse operation of the electro-optical transfer function, and wherein the periodic watermark signal is extractable from the final marked image.
17. The method of claim 16 , wherein the periodic watermark signal is normalized.
18. The method of claim 16 , wherein the electro-optical transfer function applies a constant gamma correction to the original image.
19. The method of claim 16 , wherein the electro-optical transfer function applies a SMPTE 2084 standard compliant transfer function to the original image.
20. The method of claim 16 , wherein the electro-optical transfer function applies transfer function compliant with a Dolby Perceptual Quantizer to the original image.
21. The method of claim 20 , wherein the electro-optical transfer function is compliant with ITU-R Recommendation BT.2020.
22. The method of claim 16 , wherein the original image has a high dynamic range.
23. The method of claim 16 , wherein the original image has highly varying ranges of luminance.
24. The method of claim 16 , comprising generating the local weighting factor using a noise visibility function having a gradient removal factor that prevents the noise visibility function from significantly incrementing at locations of spatial gradient in the EOTF image.
25. The method of claim 16 , comprising generating the local weighting factor using a noise visibility function having a density filter that prevents the watermark from becoming visible at spatial locations with highly dense textures.
26. The method of claim 16 , wherein the periodic watermark signal comprises a watermark block having a size of M×M sections.
27. The method of claim 26 , wherein the extracted signal comprises a block of M×M sections and each of the M×M sections of the watermark block correspond to the M×M sections of the extracted signal.
28. The method of claim 27 , further comprising generating the watermark block from a plurality of pseudo random numbers.
29. The method of claim 28 , wherein the plurality of pseudo random numbers is generated using a unique input identification.
30. The method of claim 29 , wherein the unique input identification is 32 bits in length.
31. The method of claim 28 , wherein the generation of the watermark block comprises generating a N(0,1) normal distributed random number in double precision using at least one of the plurality of pseudo random numbers as a seed.
32. The method of claim 31 , wherein the generation of the watermark block comprises using the normal distributed random number to populate a watermark pattern sequentially.
33. The method of claim 26 , comprising
generating the watermark block; and
limiting a frequency band used to mix the watermark block with the extracted signal by applying low-pass filtering to the watermark block.
34. The method of claim 33 , wherein the low-pass filtering utilizes a kernel size of 3.
35. The method of claim 16 , wherein the original image is an individual frame of a digital video.
36. The method of claim 16 , wherein the original image is an individual frame of a streaming video.
37. The method of claim 36 , wherein the streaming video includes a sequence of I frames and the method is applied to two or more of the I frames.
38. The method of claim 16 , wherein the watermark extractor uses an average of two or more local noise variances for the original image to generate the extracted signal.
39. The method of claim 16 , wherein the watermark extractor comprises a Helicon filter.
40. The method of claim 16 , wherein the final image is humanly perceptible and includes a watermark having a noise pattern that resembles film grain.
41. The method of claim 16 , wherein the periodic watermark signal uniquely identifies a viewer and a time at which the final image is communicated to the viewer.
42. The method of claim 41 , wherein the original image is a frame of a streaming video to be communicated to the viewer over a telecommunications network.
43. The method of claim 42 , wherein the telecommunications network comprises the Internet.
44. The method of claim 42 , wherein the periodic watermark signal is dynamically generated and applied to the EOTF image of the frame of the streaming video immediately prior to a time of communication of the final image in the streaming video to the viewer.
45. The method of claim 16 , wherein the mixed signal is generated based on an embedding interval.
46. The method of claim 45 , wherein the final image is communicated to a receiving device over a network connection and the embedding interval is determined based on a target CPU load for a streaming server utilized to communicate the final image.
47. The method of claim 46 , wherein the receiving device is a home set-top box.