论文标题
基于运动向量的视频隐志的成本分配原理的调查
Investigation on Principles for Cost Assignment in Motion Vector-based Video Steganography
论文作者
论文摘要
运动矢量域中的成本分配仍然是视频隐志的研究重点。图像隐志的最新研究总结了许多原则，以实现成本分配并取得了良好的结果。但是，尚未完全讨论基于运动矢量的视频隐志成本分配的基本原理。首先，本文提出了运动向量域中成本分配的三个原则，包括局部最优性，块组中的非一致性和复杂性优先级。其次，根据三个原理设计了三个相应的新型实际失真函数。最后，基于所有三个原则以提高整体性能，构建了关节失真函数。实验结果表明，不仅三个独立的失真函数可以有效地抵抗相应的地分解攻击，而且最终的关节变形可以同时抵抗三个稳定的特征。此外，它可以获得良好的视觉质量和编码效率，可以应用于实际情况。
Cost assignment in the motion vector domain remains a research focus in video steganography. Recent studies in image steganography have summarized many principles for cost assignment and achieved good results. But the basic principles for cost assignment in motion vector-based video steganography have not been fully discussed yet. Firstly, this paper proposes three principles for cost assignment in the motion vector domain, including the principle of local optimality, non-consistency in the block group, and complexity priority. Secondly, three corresponding novel practical distortion functions were designed according to the three principles. Finally, a joint distortion function is constructed based on all three principles to increase overall performance. The experimental results show that not only the three independent distortion functions can effectively resist the corresponding steganalysis attacks, but the final joint distortion can resist the three steganalysis features simultaneously. In addition, it can obtain good visual quality and coding efficiency, which can be applied to practical scenarios.
