Research On Robust Video Watermarking Algorithm Based On Compressed Domain

With the rapid development and popularization of mobile terminals and the Internet,more and more people obtain information in the form of videos through various APPs on mobile terminals.At the same time,various information security issues also arise.Pirated movies and private videos are spreading freely on the Internet,and the resolution of various copyright and information security issues is imminent.In response to these problems,video watermarking technology has emerged as an effective means to solve copyright issues and protect information security.Video watermarking technology achieve the purpose of information protection by embedding the copyright information in the video.After the video spread on the Internet,the copyright information can still be proposed from the video to prove the ownership of the video,or to trace the source of the video.However,video is subject to two common attacks during network transmission:(1)Recompression attack: In the transmission of video,due to the limitation of server bandwidth,the video generally needs to be compressed and transmitted again under the video coding standard.In this process,most users will perform lossy compression on the video by increasing the quantization parameter QP in the compression encoding to reduce the video resolution,which will affect the quality of the video and the embedded watermark;(2)Geometric attack:Because the video will change the aspect ratio after being transmitted on different players,or be recorded and broadcast by the mobile phone.In this process,the video suffers from geometric attacks such as rotation or scaling,which will affect the quality of the video itself and cause the embedded authentication information to be unable to be accurately extracted.In real-time applications,how to ensure that the watermark embedded in the video can still maintain robustness under these two types of attacks is an issue worthy of attention.In response to these two attacks and the current research status,this paper proposes two robust video watermarking algorithms based on the compressed domain.In response to recompression attacks,the paper proposes a robust video watermarking algorithm in the compression domain that can resist recompression attacks with increased quantization parameters.In the proposed method,the last P frame of each GOP of the video is selected to embed the watermark to reduce drift distortion.In this frame,the watermark is embedded in the macroblocks that are more robust to the recompression attack based on the video content,so as to improve the performance of the watermark recompression attacks.Experiments show that the proposed method can withstand the recompression attack with a large increase in QP.While reducing drift distortion,it effectively limits the degradation of video perception quality.In addition,since there is no need to record the watermark position information when the watermark is embedded,the application scenarios of the algorithm are more extensive.In response to geometric attacks,the paper proposes a robust video watermarking algorithm in the compressed domain that can resist geometric attacks.This algorithm selects areas that are more robust to geometric attacks to embed watermarks in the salient areas of each frame of the video,and improve the performance of resisting geometric attacks.When extracting the watermark,the attacked video is geometrically corrected by statistical feature points before extracting the watermark.Experiments show that the proposed algorithm is robust to geometric attacks such as rotation and scaling,and does not affect the visual perception quality of the video and does not increase the bit rate of the video.And because the algorithm embedding the watermark is implemented in the compressed domain,the computational complexity of the watermark embedding is reduced.