一种基于DCT域的视频水印算法  

An Algorithm of Video Watermark Based on DCT Domains

作　　者：郑鹏[1,2] 刘敏忠[2] 

机构地区：[1]空天信息安全与可信计算教育部重点实验室,武汉430072 [2]武汉大学计算机学院,武汉430072

出　　处：《武汉理工大学学报》2009年第18期9-11,50,共4页Journal of Wuhan University of Technology

基　　金：武汉市科技攻关项目(200810321134)

摘　　要：为了对现有的大量MPEG-2视频文件进行版权保护,需要一个能对现有视频进行快速水印嵌入,并且不影响视频图像质量的水印算法。根据这一目的,提出了一种基于MPEG-2压缩视频DCT域的有效视频水印算法。直接利用I-帧图像中的I型宏块的中频DCT系数进行水印嵌入,无须对压缩视频进行编码和解码,水印嵌入速度快,可实现水印的实时嵌入和提取。中频系数对人眼不太敏感,图像失真度小,同时选择中频系数也不会降低游程编码的性能。对MPEG-2视频进行了实验,实验结果表明方法非常有效。In order to protect the copyright of the large amount of existing MPEG-2 video files, we need a video watermark algorithm that can embed watermark quickly, and not distort the quality of the video. Based on the purpose, we present an efficient algorithm to embed watermark in DCT coefficients of MPEG compressed video in the paper. We make use of the medium-frequency DCT coefficients of Ⅰ-MBs in Ⅰ-frames to embed watermark, needless encoding and decoding the compressed video. The speed of embedding is so fast that we can realize real time embedding and extracting watermark. The medium-frequency DCT coefficients is not sensitive for people, the distortion of picture is very small, and they will scarcely any reduce the efficiency of run-level encoding. The experimental result shows that our algorithm is very efficient.

关 键 词：水印 DCT系数 MPEG 

分 类 号：TN911.73[电子电信—通信与信息系统]