A Video Watermarking Algorithm Research Based On 3D-HEVC

3D video is widely favored because it can provide real visual experience.The convenient Internet and mature multimedia technology have successfully promoted multimedia applications.While enjoying multimedia services,users also face copyright disputes caused by illegal copying,falsification,and tampering,which has brought a huge test to the commercial development of3 D video.Video watermarking technology can better solve this kind of problems,which has attracted the attention of scholars and businesses,and has become a current research hotspot.Because 3D video data is abnormally large,in order to reduce the demand for bandwidth and storage,the High Efficient Video Coding(HEVC)standard has been extended again to obtain the 3D-HEVC coding standard.The addition of multiple new encoding methods gives 3D-HEVC a considerable compression ratio.However,while the redundant information is effectively reduced,the coding complexity is also increased accordingly,which makes the design of video watermarking algorithms more difficult.The existing image and video watermarking of the algorithm cannot be effectively used in 3D video.Aiming at the problems faced by 3D video and the encoding characteristics of the new standard,two video watermarking algorithms are proposed:(1)Zero-watermarking algorithm for anti-recoding in depth image frames.After experimental tests,the pixel difference in the edge area of the depth map has strong anti-recoding ability.According to this feature,first,the depth map is divided into 64 × 64 blocks without overlapping;Second,each block is divided into two regions using four methods: up and down,left and right,top left and bottom right,bottom left and top right.Calculate the gradient value of each method based on the average of the pixels in the two regions;Finally,compare the maximum gradient value and threshold in four ways,and map the feature sequence.,which is XORed with the chaotic scrambled test picture togenerate a zero watermark.Under the different QPs(Quantization Parameters)(25,30,35,40)re-encoding,before and after the five test sequences were attacked,the similarity of the watermark sequence was above 92%,which proves that the anti-recoding ability is strong,and it can also better resist common noise attacks in video transmission.(2)Video watermarking algorithm based on texture complexity and motion direction.First,after statistical analysis of I-frames using the gray level co-occurrence matrix and gradient matrix,the analysis results are used as thresholds to select 8 × 8 CUs with higher texture complexity in P and B-frames.The embedded block of the watermark is combined with the motion vector component size of the co-located texture block to determine the position of the embedded watermark.Finally,the watermark is embedded by modulating the search range of the current block motion vector or modifying the size of the region's optimal motion vector.The average embedding capacity of this algorithm for different video sequences reaches 613(bit /frame).The average bit error rate under recoding attacks with different QPs(25,30,35,40)values is 13.31%,and the average drop in PSNR value is only0.0075 d B,which has almost no effect on the quality of the synthesized viewpoint,and has good robustness.