Psycho-visual-model based video watermark gain adaptation
A method for embedding video watermarks. Areas of poor visual quality in a video content having embedded watermarks ar determined. The watermark symbols replace pixels in the video content with pixels in which the luma values are modulated such the luma value for a 0 bit renders as black and the luma value for a 1 bit renders a shade of gray. The luma value for the 1 bit level is reduced in those areas determined to have poor visual quality. Areas of poor visual quality may be determined by analyzing the luma values of pixels of the video content immediately adjacent to and below a particular watermark symbol to derive an adjacent brightness parameter that represents the perceptual brightness of the surrounding video content. The reduction of the luma value may be achieved by setting the 1 bit luma for a particular watermark symbol to be perceptually less bright than the adjacent brightness parameter.
1. A method comprising:
determining areas of poor visual quality in a video content that may be the result of watermark embedding by analyzing the luma values of pixels of the video content immediately below a particular watermark symbol to derive a brightness parameter that represents the perceptual brightness of the surrounding video content;
determining watermark robustness requirements;
embedding the video content with a watermark including watermark symbols, wherein the watermark symbols replace pixels in the video content with pixels in which luma values are modulated such that the luma value for a 0 bit (“Bit0”) renders as black and the luma value for a 1 bit (“Bit1”) renders as a shade of gray; and
adjusting the 1 bit luma level based on visual quality and said watermark robustness requirements.
2. The method of claim 1 wherein reducing the Bit1 level comprises:
setting the Bit1 luma for a particular watermark symbol to be perceptually less bright than the brightness parameter.
3. The method of claim 2 further comprising setting Bit1 luma as a percentage of the difference between the derived parameter and the minimum value that Bit1 can take.
4. The method of claim 1 further comprising detecting the watermark by calculating a symbol detection threshold across a segment of a top line of the video.
5. The method of claim 4 further comprising changing the size of the segment based on variations in the luma signal.
6. The method of claim 1 further comprising detecting the watermark by:
analyzing the luma signal before and after a symbol boundary to determine whether there was a transition from Bit1 to Bit1, a transition from Bit1 to Bit0, or no transition, thereby enabling individual symbol detection of embedded content where each symbol might have a different Bit1 value.
7. The method of claim 1 further comprising calculating a value for adjacentBrightness using fixed segment boundaries that span several symbols, entire frame, or multiple frames, such that all Bit1s embedded within that span have the same luma value.
8. The method of claim 7 further comprising detecting the watermark using the same fixed segment boundaries to optimize the detection threshold for each segment.
9. The method of claim 1 further comprising calculating a value for adjacentBrightness using variable size segment boundaries, wherein the segment size is chosen to include only symbols where the variance of luma is below a threshold.
10. The method of claim 9 wherein the calculating further comprises calculating the running variance of the luma of an adjacent host content and creating a segment boundary when the running variance exceeds a threshold, wherein a fixed segment size is used if the variance is above the threshold.
11. The method of claim 10 further comprising detecting the watermark by calculating the running variance of the luma of an adjacent host content and locating segment boundaries wherein a detection threshold is optimized for each segment.
12. The method of claim 1 further comprising:
determining tuning parameters, including percentDimmer, bit1Min, bit1Nominal, minSpreadBit0, and minAdditiveDifference.
13. The method of claim 12 , further comprising setting the tuning parameters to fixed values.
14. The method of claim 12 further comprising varying the tuning parameters symbol by symbol.
15. The method of claim 12 further comprising varying the tuning parameters frame by frame.
16. The method of claim 15 further comprising improving detectability a high priority frame by selecting the tuning parameters for the high priority frame for optimal detectability at the expense of visual quality for that frame.
17. The method of claim 12 further comprising:
evaluating in real time the detectability of the embedded frame during watermark embedding by processing the embedded frames and running a detector on those processed frames;
calculating a detection error metric based on the detection; and
modulating at least one of the tuning parameters so as to maintain watermark robustness at a desired level.
18. The method of claim 1 further comprising prior to determining areas of poor visual quality in a video content, detecting the presence of existing watermarks.
19. The method of claim 1 wherein watermark robustness requirements are determined based on at least one of the one of the following considerations:
a. whether the symbol belongs to an important message;
b. whether the message includes FEC;
c. whether embedded symbols are found experimentally to lack robustness;
d. the video entropy of the area;
e. a measurement of video motion;
f. expected processing in a content broadcast path; and
g. expected perceptual compression algorithms.