摘要
在现有视频水印算法的基础上,结合H.264编码标准的新特性,提出一种基于H.264压缩域的鲁棒性视频水印算法.利用子块的纹理复杂性,将I4宏块中有较多非零量化系数的4×4子块作为嵌入区域,针对已有算法中比特率增加这一问题,该算法利用不同帧内预测模式下DCT系数的分布特点确定概率矩阵,据此通过改变子块的非零量化系数个数的奇偶性将水印嵌入到每个I帧亮度块量化后的DCT系数中.仿真结果表明,该算法对视频码率的影响较小,在保证视频水印的不可见性和鲁棒性的前提下,使得水印嵌入容量有了较大提高,同时该算法能够满足实时盲检测的要求.
A robust video watermarking algorithm in H. 264 compressed-domain is proposed based on the new characteristics of H. 264 coding standard. The 4 x 4 blocks which have more nonzero quantized coefficients in I4 macroblocks are chosen as embedding locations based on their texture complexity. To overcome the short- coming of the bit-rate increase in the existing algorithm, probability matrix is determined according to distribu- tion characteristics of DCT coefficients under the different intrapredicted models. The watermark is embedded into the quantized DCT coefficients of the luma components in I frames by changing the odevity of the number of the nonzero quantized coefficients. The simulation results show that the proposed algorithm has little impact on the bit-rate of video, good invisibility and robustness. It can also increase watermarking embedding capaci- ty and meet the requirements of real-time blind detection.
出处
《郑州大学学报（工学版）》
CAS
北大核心
2013年第5期63-67,共5页
Journal of Zhengzhou University（Engineering Science）
基金
河南省基础与前沿技术研究计划资助项目(102300410003)
关键词
H
264
视频水印
量化系数
盲检测
H. 264
video watermarking
quantized coefficient
blind detection
