Research On H.264/AVC Video Watermarking Algorithm

Today, with the internet product penetrating into people’s live, digital video, as animportant part of the internet product, will have have a greater impact in people’slives. However, as an inherent defect of digital products, easily to be modified is afate that the digital video can not avoid. Therefore, video product’s copyright, legalityand its reliability in contents are difficult to be secured. Fortunately, the inherentdefects of digital video products can be covered by video watermarking. Therefore,the study of video watermarking technology will become more in line with thedemand of the market. H.264/AVC compression standards, as a new generation ofvideo compression, have a better compression performance than previouscompression standards. Because of this, it is widely recognized by the market.Therefore, the research of video watermarking technology based on H.264/AVC has abroad market prospect.On this basis, three video watermarking algorithms based on the H.264/AVCcompression standard were presented in this paper:1，The video watermarking algorithm is based on inter-frame sub-macroblockmode selecting. In order to embed watermark into the video, the parity of thesub-block’s number was modified indirectly when the mode of sub-macroblock wasmodified according to the watermark. In order to combine with HVS, thosesub-macroblocks whose SAD values have a minimum change after modification wasselected as the embedded region. Simulation results have shown that the proposedalgorithm has good robustness against noise attacking. Meanwhile, it has a smallinfluence on the bit rate, and also has a good effect in the PSNR of video. Theextraction algorithm is a blind extraction.2，The reversible video watermarking is based on DCT coefficients. Combinedwith HVS, we count up the number of non-zero coefficients at first. Meanwhile, the4×4sub-block who has the most non-zero coefficients in one macroblock wasselected as the embedded region. Then,4×4sub-blocks are classified according to the number of nonzero coefficients. For different categories of4×4sub-blocks, we usedifferent strategies to embed watermark. At last, according to the value of watermarkdata, the last non-zero coefficient and the followed zero coefficient was modified soas to implement the watermark embedded. Simulation results have shown that it has asmall influence on the bit rate and video quality, and it also can restore the videocompletely. The extraction algorithm is a blind extraction.3，The last fragile video watermarking which is used for solving the problem ofintegrity authentication for H.264/AVC video was presented on the inter-frame mode,motion vector and the high frequency DCT coefficients of I frame. In each of theGOP, the feature code was created by every macroblock’s best mode and motionvector in each P frame. At the same time, in order to get the fixed watermark datawhich will be embedded in video, we transform this feature code with hash functionwhen one GOP is going to end. Then, the reversible watermarking algorithm wasused to embed the fixed watermark data into I frame which is in the next GOP. The4×4sub-block as embedded area was determined by the4bits of feature code beforethe current watermark bit. In the decoder, we can extract the watermark data correctlybecause the reversible watermarking algorithm can restore the I frames completely,and the error can’t spread into P frames. Simulation results have shown that it has asmall influence on the bit rate, almost no difference to the quality of video. At thesame time, its fragility is very well, and it can realize video integrity authentication inthe macroblock level. The extraction algorithm is a blind extraction.