随着数字视频的普及,HEVC(high efficiency video coding)视频隐写受到了越来越多的关注。残差系数域隐写具有高安全性和低码率增长的特点,然而,视频编码过程中重构的系数不满足块间相互独立的性质,导致修改视频残差系数往往会带来较为严重的失真漂移。本文提出了一种无帧内失真漂移的HEVC视频自适应隐写算法。首先,根据帧内预测方向和多系数修改的方法选择合适的载体,进而避免修改系数引起的失真漂移。然后,设计一种包含块内失真和修改载体后的系数分布的失真函数,指导校验网格编码(syndrome trellis codes,STC)对嵌入失真小的载体进行修改。最后,根据最小化失真代价将消息嵌入到满足特定条件的4×4亮度块的量化正弦变换(quantized discrete sinc transform,QDST) 系数中。实验结果表明,所提出的视频隐写算法能够有效地避免帧内失真漂移,在获得良好的视觉感知质量的同时,能保证隐写算法的安全性。

With the popularity of digital video,high efficiency video coding (HEVC) video steganography has received more and more attention.Residual coefficient domain steganography has high security and low bit rate increase.However,during the process of video encoding,the reconstructed coefficient does not meet the independence between blocks,resulting in the modification of video residual coefficients often lead to more serious distortion drift.In this paper,an adaptive steganography algorithm for HEVC video without intra frame distortion drift is proposed.Firstly,the appropriate carrier is selected according to the intra frame prediction modes and the multi-coefficient modification method so as to avoid the distortion drift caused by the coefficient modification.Secondly,the distortion function is designed which contains two factors,i.e.,intra block distortion and modified carrier coefficient distribution.It is employed to guide the syndrome trellis codes (STC) to modify the carrier with low embedded distortion.Finally,according to the minimum distortion cost,the message is embedded into quantized discrete sine transform (QDST) coefficient of the 4×4 luminance block,which meets specific conditions.Experimental results show that the proposed video steganography algorithm can effectively avoid intra frame distortion drift and ensure the security of the steganography algorithm while achieving good visual quality.

国家自然科学基金项目(62071267) 和浙江省自然科学基金项目(LR20F020001) 资助项目