Video Watermarking Algorithm In H.264 Compressed Domain

In this information age,with the popularization and application of the Internet,digital information products have become the main media for contemporary information transmission,and digital video has become the most widely and easily accepted information transmission format among these media.Accompanying more and more mature video technology are the copyright issues and information security issues encountered in the process of copying,forwarding,and tampering of these digital video information.To solve these problems,video watermarking technology has been proposed.Video watermarking technology can hide some important information,such as copyright information,in the video,not only can it meet the invisibility and robustness of important information,but also it can spread the information along with the video and ensure that it can also be obtained at the receiving end.Embedded watermark information.Traditional video watermarking algorithms are based on the pixel domain.The advantage of this is that the algorithm design is simple and the watermarking algorithm is more robust.However,the disadvantage is that the amount of calculation is large,and embedding and extracting watermarks also require complete encoding and decoding of the video.Such an inefficient solution does not meet the requirements of real-time watermarking.In order to reduce the calculation amount of the watermarking algorithm and improve the efficiency of the watermarking system,video watermarking algorithm based on compression domain has gradually become a research hotspot.Because video encoding is lossy compression,every time the video is encoded,the internal data of the video will be irreversibly changed,causing the information embedded in the video to be lost to varying degrees.Compression attack.When the problem of anti-recompression of video watermarks is actually considered,the quantization step size parameter QP is also a parameter that cannot be ignored.QP is used in video encoding and decoding,and affects the quality and bit rate of the video.The robustness of video watermarking under heavy compression attacks is also the most important research.In view of the current series of problems facing video watermarking,we combined the most widelyused video codec H.264 / AVC,using video codewords and discrete cosine transform coefficients,and performed video watermarking algorithms on JM16.0.Further research.The main contents of this article are as follows:(1)A saliency area detection algorithm based on H.264 / AVC compression domain was implemented.This algorithm mainly uses two feature factors of motion vector entropy(MVE)and quantized transform residual norm(SRN)to evaluate the saliency of a macroblock.The two feature factors are used to calculate the significance of a video frame image Sex area.Through our experiments,we found that the codec will retain as much information as possible when encoding the saliency area,so it will also assign more encoding parameters to the saliency area.Therefore,embedding the watermark into the saliency area can improve the robustness of the watermark to a certain extent.Therefore,the saliency algorithm lays the foundation for the subsequent watermarking algorithm.(2)Improved a codeword replacement watermarking algorithm based on compressed domain.The original algorithm is to replace the random macroblock in the video frame image with the codeword obtained after entropy coding.The advantage of this algorithm is that the concealment of the watermark is high and the impact on the video is small.But the biggest disadvantage is that it is less robust and cannot resist re-compression attacks,especially re-compression attacks with fixed quantization steps.To solve this problem,we combined the theoretical analysis to improve the watermark embedding position in the original algorithm.We first use the saliency region detection algorithm to filter out the saliency regions in the video,and then embed the watermark into the saliency regions.The experimental results show that the effect of the algorithm on the original video is reduced by 30% compared with the original algorithm.In particular,the improved algorithm also has anti-recompression performance.After recompression,the watermark error rate is less than 10%.(3)A video watermarking algorithm based on compression domain is designed.In order for the watermark algorithm to be applied in practical applications,we must make the watermark resistant to recompression attacks with non-fixed quantization steps.Because modifying the quantization step size QP will cause the overall video data to change,the traditional watermarking algorithm cannot resist quantization attacks.In this algorithm,we combine two concepts of feature factors: residual factor and energy factor.By fusing thesetwo factors and correcting the motion vector entropy in the saliency area algorithm to correct them,some relatively stable macroblocks can be filtered out.These macroblocks can be accurately found through the above methods under different quantization steps.Finally,the watermark embedding method designed by us can be used to completely embed the watermark into the video.Experiments show that the algorithm can well resist re-compression attacks and quantization attacks.The bit error rate of the watermark after re-compression and quantization attacks does not exceed 15%,and the impact of embedded watermarks on video is very small.In particular,the algorithm has no effect on the bit rate of the video.