Y. Cao, X. F. Zhao, D. G. Feng, R. N. Sheng
In this paper, we propose an adaptive video steganography tightly bound to video compression. Unlike traditional approaches utilizing spatial/transformed domain of images or raw videos which are vulnerable to certain existing steganalyzers, our approach targets the internal dynamics of video compression. Inspired by Fridrich et al’s perturbed quantization (PQ) steganography, a technique called perturbed motion estimation (PME) is introduced to perform motion estimation and message hiding in one step. Intending to minimize the embedding impacts, the perturbations are optimized with the hope that these perturbations will be confused with normal estimation deviations. Experimental results show that, satisfactory levels of visual quality and security are achieved with adequate payloads.
Cite this paper as:
 Y. Cao, X. F. Zhao, D. G. Feng, R. N. Sheng. Video steganography with perturbed motion estimation, In Revised Selected Paper of 13th International Workshop on Information Hiding (IH 2011), Prague, Czech, May 18-20, 2011, LNCS 6958, pp. 193-207, Springer. [PAPER]