Watermarking Research Based On H.264/AVC Video

The rapid development of information technology makes it very convenient to acquire, copy,edit, and transmission the digital video. However, the case of illegally using of digital video arebecoming an increasingly serious problem, leading the information security becoming more andmore prominent. Digital watermarking technology, being studied and as an effective solution toprevent the digital video being pirated, counterfeit, altered and attacked, has attracted the generalconcern of scholars, and gradual commercialized. Digital watermarking is the process ofembedding copyright information such as author/owner/usage restrictions into the host dataimperceptibly and robustly. Watermarking is an ideal solution for protection of intellectual propertyrights. However, as the video is usually stored and transmited as the compressed format, H.264asthe most efficient and the latest compression standard is utilized in a wide range of application,therefore, providing a secure information hiding method, which is appropriate for this standard, ishighly desirable.This paper illustrated the background of the video watermarking and the situation of theresearch firstly, then we discussed the model of the video watermarking, the features, theclassification, the application and the evaluation method of its performance, on this basis, thewatermarking method of our research is illustrated.Firstly, we made a statistical analysis of the H.264encoder, such as the symtax elementsincluding the residual coefficient, the motion vector difference, the encode mode and the predictionmode, etc. we utilized its statistical features to design varaties of watermarking methods ofdifferent performances, furthermore, this paper proposed kinks of method to solve the problem ofthe video watermarking, such as how can we emhance the robustness to resist the re-encode attack,the desynchronization problem, how can we increase the capacity of the watermarking, etc.Secondly, according to the charictiristic of the DCT coefficient, this paper proposed two kindsof method to emhance the robustness to resist re-encode attack, index them A and B. Modulatingthe DCT coefficients of Intra₄4and Intra₁616in I-frames to embed the watermark. InIntra₄4, the44sub-block which has the biggest number of non-zero coefficients is selected asthe embedding target, method A modified its high-frequency DCT coefficient to inseart thewaterking, the strength of embedding is determined by the quantization parameter, while method Bfirstly categoried two group accorking to the DCT coefficient, then modified the higher-frequencyDCT coefficient to map the sub-block to corresponding group according to the watermark. InIntra₁616, the two methods are all inseart the watermark by moditying the LSB bit of the DCluma coefficient. The watermark can be extracted just by decoding each I frame and do not in needof the original video, which meets the requirements of the modern detection of real-time perfectly.Simulation results show that the algorithm brings little effect on distortion and bit-rate increase to the video, and has a good performance in the resistance to the recoding.Thiredly, a large capacity of data hiding algorithm was proposed based on the motion vectorin P and B slices, with lower bit-rate can be achieved in the mean while. The main idea is to hidethe information into appropriately selected motion vectors in Inter frames, we modulate LSB bit ofthe x and y sub-components of each MVD data to hide2bits information, in this way we canexploit the large hiding capacity as well as the lower bit-rate increase. The data can be extractedjust by decoding part of the motion vector difference and do not need the original video stream,which meets the requirement of the modern detection of real-time perfectly. Simulation resultsshow that the algorithm can achieve a great embedding capability as well as the slight bit-rateincrease and visual distortion.Furthermore, this paper proposed a method to enhance the security for the data hidingalgorithm, as the importance of the secret information, the detector should make an effectivemethod to determine the reliability of the extracted information, we embed the watermarking intothe video which the secret data had hided in it, in this way we can authenticate the integrity of thevideo stream. Furthermore, by the detected watermark and its integrity, the receiver can determinethe reliability of the extracted hided information and judge whether the video has suffered fromvarious attacks in addition, the security of the secrete communication can be guaranteed.Finally, a content authentication watermarking algorithm based on the CABAC entropycoding for the H.264/AVC is proposed. The specific motion vector difference in P and B frames areselected according to the characteristic of the CABAC. The watermarks are embedded bydisplacing the binary sequence of the selected syntax elements regularly. Experimental resultsdemonstrate that the algorithm can achieve a great embedding capability as well as the slightbit-rate increase and visual distortion. The embedding and extraction processes can performquickly, simply and real-time. It’s appropriate to certificate the integrity of the video as thevulnerability of the algorithn.