基于帧内预测模式多划分的HEVC鲁棒视频水印算法

发布时间：2018-05-27 14:33

本文选题：高效视频编码 + 帧内预测模式　；参考：《计算机应用》2017年06期

【摘要】：针对现有基于高效视频编码(HEVC)标准的水印算法鲁棒性不足的问题,提出一种基于帧内预测模式多划分的HEVC鲁棒视频水印算法。首先,针对嵌入水印后帧内误差传播的问题,对4×4亮度块进行可嵌区域的选择,并计算4×4亮度块的纹理方向;其次,将33种角度预测模式划分为四种模式集,依次记为:上水平、下水平、上垂直、下垂直;最后,将当前以及下一待嵌入水印的值与模式集建立映射关系,通过判断当前4×4块属于哪个模式集来进行水印的嵌入,并将33种角度模式截断为其中一种模式集。解码端通过纹理方向和预测模式集提取水印。实验结果表明,所提算法的平均峰值信噪比维持不变,且在重编码攻击下的误码率(BER)为14.1%。由此可知,该算法对视频质量影响很小,在鲁棒性上可以抵抗重编码的攻击。
[Abstract]:Aiming at the lack of robustness of the existing watermarking algorithms based on the efficient video coding (HEVC) standard, a robust video watermarking algorithm based on intra prediction mode multi-partition (HEVC) is proposed in this paper. Firstly, aiming at the problem of intra-frame error propagation after embedding watermark, 4 脳 4 luminance block is selected and the texture direction of 4 脳 4 luminance block is calculated. Secondly, 33 angle prediction modes are divided into four pattern sets. In turn, it is recorded as: upper level, lower level, upper vertical, lower vertical; finally, the value of the current and the next watermark to be embedded is mapped to the pattern set, and the watermark is embedded by determining which pattern set the current 4 脳 4 block belongs to. The 33 angle patterns are truncated into one of them. The decoder extracts the watermark by texture direction and prediction pattern set. The experimental results show that the average PSNR of the proposed algorithm remains unchanged, and the bit error rate (BER) of the proposed algorithm is 14.1 under the re-coding attack. Therefore, the algorithm has little effect on video quality and can resist recoding attack in robustness.
【作者单位】：华侨大学信息科学与工程学院;
【基金】：福建省自然科学基金资助项目(2016J01306) 华侨大学研究生科研创新能力培育计划资助项目(1511301038)~~
【分类号】：TN919.81;TP309.7

【相似文献】