Cross correlation feature mining for steganalysis of hash based least significant bit substitution video steganography

摘要

Recently, video steganography has attracted researchers' attentions for its relative higher capacity than steganographic technique using image or audio as cover. Various image steganographic techniques have been proposed, such as least significant bit (LSB) substitution, LSB matching, etc. In (Dasgupta et al. in Int J Secur Priv Trust Manag 4:1-11, 2012), the authors proposed a hash based least significant bit substitution video steganography in spatial domain, and eight bits of secret message were divided into 3, 3, 2 segments and embedded into the RGB pixel values of cover frame. Besides, a hash function was used to determine the embedding locations. However, the hash function is fragile to attack and the observer can easily figure out the exact embedded positions. Moreover, the embedding of secret message significantly changes the cross correlation feature of consecutive video frames. Based on cross correlation analysis, a video steganalysis technique is proposed in this paper. Theoretical analysis and experimental results all show the effectiveness of the proposed cross correlation based video steganalysis technique.