Research On Steganography Based On Intra-prediction For H.264 Video

With the popularization of the Internet,the degree of informatization has been greatly improved,and the frequent occurrence of security problems has drawn people's attention.For video steganography,steganography based on H.264 / AVC is a hot topic of network security.Moreover,scholars have done a great deal of work and made a lot of achievements.In our country,video steganography has also been strongly support by the state.Video steganography has been widely used in military espionage because of its strong covertness and high embedded capacity.However,as video steganalysis tools become more powerful,many steganographic schemes cannot resist these tools.Therefore,my thesis will discusses new ideas and focus on the research based on the framework of minimized distortion,and propose a steganalysis scheme with higher security and coding efficiency.The main research and work of this paper focus on the construction of the cost function and discuss it with the H.264 compression coding.The researches is as follows:(1)A video steganography scheme is proposed to effectively resist the calibration analysis,and it combines the minimization distortion embedding framework and takes advantage of the inherent characteristics of the intra prediction mode in the H.264 / AVC encoding format.Taking the sum of absolute different(SAD)generated by intra prediction as the key factor in constructing the cost function.At the same time,the mapping rule based on the prediction deviation is introduced,so that the SAD deviation generated by modifying the mode is smaller.(2)A video steganography scheme is proposed to effectively resist the spatial correlation analysis.This scheme focuses on constructing the cost function.By using the characteristics that the spatial correlation of the texture edges is hard to model,the edge feature is added to the cost function,making embedding as close as possible to the edge area that is difficult to detect.After combining the edge-based mapping rules,the prediction direction difference between the optimal prediction mode and the candidate mode is reduced.The experimental results show that the proposed scheme shows good security against spatial correlation analysis and guarantees the coding efficiency and visual quality.(3)Compared with the existing steganography algorithms based on intra prediction,the steganographic algorithm based on calibration analysis design achieves higher security and also controls the encoding efficiency while ensuring visual qualit.Moreover,the video steganography scheme that resists spatial correlation analysis also shows better security while ensuring high coding efficiency and visual quality.