一种基于量化索引调制的视频盲水印算法

针对传统视频水印算法鲁棒性较差、计算复杂度较高的问题，结合伪三维DCT（discrete cosine transform）变换，提出了一种新的基于量化索引调制的非压缩域视频盲水印方法。该算法随机选择帧提取亮度分量，进行伪三维DCT变换，并计算交流分量的权值和，利用量化索引调制与人类视觉系统特性将水印嵌入视频，还生成了一个随机密钥，有效提高了视频水印的安全性。提取水印时不需要原始视频序列，减少了算法运行时间。实验结果表明，该算法对视频质量的影响较小，并且能有效抵御旋转、缩放、剪切、加噪、滤波、帧交换、帧删除及帧平均等多种攻击，具有较强的鲁棒性。

Video blind watermarking algorithm based on quantization index modulation

Due to some drawbacks of traditional video watermarking algorithms, such as bad robustness and high complicated computation, a new video blind watermarking algorithm is proposed based on the quantization index modulation and pseudo 3-D DCT transformation. The algorithm randomly selects some frames in video to extract the luminance component, makes a pseudo 3-D transformation, and computes the weighted sum of AC components. Then the watermark is embedded into video by utilizing the quantization index modulation and human vision characteristics. Meanwhile, a random secret key is produced to enhance watermark security. For more generality and less computation time, the original video is not necessary for watermark detection and extraction. The simulation results show the algorithm has little influence on the video quality, and effectively resists multiple attacks, such as rotation, zoom, cropping, noise, filter, frame exchange, frame deletion and frame averaging. In a word, the new algorithm has better robustness and less computation burdens.