Application of list decoding of convolutional codes in steganography

摘要

Convolutional codes with Viterbi decoder are widely used in content-adaptive steganography. Although, this approach improves imperceptibility of stego image, it has a substantual drawback. Viterbi algorithm allows only additive metric to be optimized, while steganalysis can exploit this property. In this paper, we introduce the solution of the problem with ML list decoding procedure. We propose to obtain list of possible vectors by additive criterion and then select one of them by steganalysis criterion.