菜单

Motion Vector-Based Video Steganography with Preserved Local Optimality (MTA, 2016)

02/03/2018 - Video Steganology

H. Zhang, Y. Cao, X. F. Zhao

ABSTRACT

Current motion vector based video steganography is unable to preserve the local optimality of modified motion vectors. Thus they are vulnerable to the attack of steganalysis. In this paper, we have proposed a novel method to guarantee the local optimality of modified motion vectors. To modify a motion vector, firstly designate a search area which consists of candidate motion vectors. Second, evaluate the local optimality of each motion vector in the search area to locate all local optimum ones, from which finally select the one contributing least to video compression efficiency degradation as the modified motion vector. Highly undetectable motion vector based video steganography can be developed by combining the proposed method with steganographic codes and reasonable cost assignment. Comparative experimental results have demonstrated that video steganography based on the proposed method is capable of withstanding current best steganalysis while keeping the video compression performance.

Cite this paper as:

[1] H. Zhang, Y. Cao, X. F. Zhao. Motion vector-based video steganography with preserved local optimality. Multimedia Tools and Applications, 75(21): 13503–13519, 2016. [PAPER] [C_WIN]