Steganalysis by Codeword Error Bits Run-length against Coding Channel Steganography

Coding channel steganography is a kind of approach that utilizes the redundancy brought by channel coding to perform reversible information hiding. However, steganalysis against coding layer steganography is still a pressing problem. This paper proposes a coding channel steganalysis scheme based on the run-length of error bit number in codewords. By counting the number of error bits in each codeword, the distribution of different lengths of runs can be extracted as features. Support Vector Machine is used as the classifier. Experimental works, which are divided into two parts, are presented to prove that the proposed scheme significantly outperforms prior to traditional image-based steganalysis methods which fail to perform well in detecting hidden information in the coding layer. Particularly, we discuss the attack effect of the algorithm against three steganographic algorithms in two cases and analyze the experimental results from different perspectives.