Using the generalised Viterbi algorithm to achieve a highly effective stegosystem for images

Abstract

The HUGO project, published at 2010, can be considered as one of the most promising direction in the design of highly undetectable steganography. The main idea of that approach is to minimise the embedding impact from the steganalysis point of view. This goal is achieved by using trellis codes in the embedding procedure, the Viterbi algorithm (VA) and the SPAM features. But the optimality of VA was kept still unclear because a generic purpose of VA is to correct errors with trellis codes instead of embedding secret information. The first goal of the current paper is to prove the optimality of VA application in its generalised form proposed about 30 years ago by one of the authors of this paper. The second goal is to optimise the parameters of the trellis code check matrix for better undetectability of stegosystems.