利用视频帧块的高效半盲鲁棒水印算法

Robust Semi-blind Watermarking Scheme of Video Frame Block

针对视频水印算法鲁棒性差或需过多先验信息提高鲁棒性的问题,提出利用视频帧块、KAZE特征匹配和DCT变换方法实现半盲鲁棒视频水印算法。首先,利用用户注册信息作为水印并为之产生唯一的秘钥,基于该秘钥随机产生唯一的帧块并存入数据库。然后,将帧与帧块进行KAZE特征匹配并检测出嵌入区域,进而嵌入水印。最后,将帧块与扭曲视频进行KAZE匹配修正视频各帧并提取水印。仿真试验结果显示,在半盲的前提下,本算法具有高鲁棒性、水印提取过程同步性、版权纠纷解决和非法传播追踪的优点,且具有高应用价值。

For the problem of Robust or need much information for realizing Robust of video watermarking scheme, the scheme is proposed that realizes the semi-blind robust video watermarking with video patch, KAZE feature matching and DCT transform. Firstly, user register their information and make an unique key for each one, with the unique key randomly produce the unique flame-patch and save the frame-patch to database. Secondly, flame and frame-patch do KAZE feature matching and detect the embedded region. Lastly, flame patch and the distorted video do KAZE feature matching and extract the watermarking. Experimen- tal results show that the proposed method is robust, synchronization of extraction watermark process and fix the problem of digital property dispute and illegal user tracing