基于运动矢量的视频隐写中代价分配原则研究
Investigation on Principles for Cost Assignment in Motion Vector-based Video Steganography
AI总结：
本文提出运动矢量域视频隐写的三项代价分配原则，设计相应失真函数并构建联合失真函数，实验证明其能有效抵抗多种隐写分析攻击且保持良好视觉质量与编码效率。
摘要：
运动矢量域中的代价分配仍然是视频隐写的研究重点。近年来，图像隐写领域的研究总结了许多代价分配原则并取得了良好效果，但基于运动矢量的视频隐写中代价分配的基本原则尚未得到充分讨论。本文首先提出了运动矢量域中代价分配的三项原则，包括局部最优性原则、块组内非一致性原则和复杂度优先原则。其次，根据这三项原则设计了三个相应的新颖实用失真函数。最后，基于所有三项原则构建了一个联合失真函数以提升整体性能。实验结果表明，不仅三个独立的失真函数能够有效抵抗相应的隐写分析攻击，最终的联合失真函数还能同时抵抗三种隐写分析特征。此外，该方法能获得良好的视觉质量和编码效率，可应用于实际场景。
英文摘要：
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.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
