一种量化误差补偿的压缩域H.264/AVC视频水印算法

提出一种基于H.264的压缩域视频水印方法.根据块编码模式参数CBP和非零DCT系数个数,对DCT块分类,选取非零DCT系数较多的块作为嵌入区域.为了保持嵌入前后视频流码率基本恒定,水印只嵌入到非零DCT系数.此外,提出了一种量化误差补偿机制,它对不同的DCT系数定义了一个误差准则函数,根据该函数的值动态地选择水印的嵌入位置,并根据视频的修改程度,自适应地嵌入不同强度的水印.仿真实验结果表明,该算法能快速地嵌入和提取水印信息,水印嵌入容量较大,且嵌入补偿机制有效降低了量化和水印引起的误差.

A Compressed-Domain Video Watermarking Algorithm for H.264 AVC with Quantization Error Compensation

A compressed-domain video watermark algorithm for H.264AVC is proposed in this paper.The DCT blocks are classified according to code block pattern(CBP) and the number of nonzero DCT coefficients,and those blocks with more non-zero DCT coefficients are selected for watermark embedding.To keep the bit-rate stable,watermark is only embedded into the non-zero DCT coefficients.Moreover,a compensation mechanism for quantization error is proposed.It defines an error norm function for different DCT coefficients to dynamically choose the embedding position,and adaptively embed watermark of different strength according to the degree of modification.Experimental results demonstrate that the proposed approach can embed and extract the watermark information efficiently with high embedding capacity,and the embedding compensation mechanism effectively decreases the error introduced by quantization and watermark embedding.