Research On Robust Video Watermarking Algorithms Against HEVC Compression

With the vigorous development of Internet and information technology,and the rise of various short video applications,videos have penetrated all aspects of people's lives.They enrich people's lives and improve their living standards,but they also pose serious threats to information security,especially the security of video data.Due to the lack of effective copyright protection measures,pirated videos have flooded the Internet,and various video infringement issues and copyright disputes emerge in endlessly.Copyright protection for digital videos is an issue that must be resolved urgently.Due to the good concealment,robustness,and stability,robust video watermarking provides a feasible solution to copyright protection.Aiming at problems in existing robust video watermarking algorithms,including the difficulty of realizing the optimal balance between imperceptibility and robustness,the weak resistance to high efficiency video coding(HEVC)compression,and the poor robustness to large-scale geometric attacks,this thesis studies the robust video watermarking algorithm in the transform domain.The main work done is as follows:(1)To solve the problem that existing video watermarking algorithms have weak resistance to HEVC compression,and better resolve the contradiction between imperceptibility and robustness,a hybrid transforms-based robust video zero-watermarking algorithm is designed.The zero-watermarking technology is introduced into the video watermarking algorithm to resist HEVC compression attacks.The multiresolution characteristic of the discrete wavelet transform(DWT),the better low-frequency energy gathering characteristic of the all phase biorthogonal transform(APBT),and the stability of singular value decomposition(SVD)are combined to extract robust features from video sequences,and robust zero-watermarks can be constructed.Since the designed algorithm does not modify the video,it will not have any impact on the video's quality.In addition,the watermark post-processing based on simple averaging method improves the accuracy of watermark detection,and further improves the robustness of the algorithm.Experimental results show that the algorithm is robust to HE VC compression with different quantization parameters(QPs),and can resist noise,filtering,scaling,cropping,sticking,and frame-based attacks.However,its resistance to rotation attacks,especially large-scale rotation attacks,is relatively weak.(2)Aiming at the problem that the previous algorithm has weak resistance to large-scale rotation attacks,a robust video zero-watermarking algorithm based on polar complex exponential transform(PCET)and Logistic mapping is designed.By analyzing the pixel loss of video frames after suffering different angles of rotation attacks,variable regions for feature extraction are determined for videos with different resolutions.The PCET with geometric invariance is introduced into the video zero-watermarking algorithm,and combined with Logistic mapping,feature sequences can be extracted from videos.The watermark image is encrypted by Fibonacci transform.The exclusive OR(XOR)operation is performed on the encrypted watermark and the extracted feature sequence to obtain the zero-watermark.The majority voting strategy is introduced into the watermark post-processing process,which improves the accuracy of watermark reconstruction.Experimental results indicate that the designed algorithm has good resistance to rotation attacks,and can also resist HEVC compression.In addition,the algorithm has high robustness to noise,filtering,scaling,cropping,frame-based attacks,and hybrid attacks.