Textural Features for Steganalysis

摘要: It is observed that the co-occurrence matrix, one kind of textural features proposed by Haralick et al., has played a very critical role in steganalysis. On other hand, data hidden image texture area been known difficult to detect for years, and modern steganographic schemes tend embed into complicated where statistical modeling becomes difficult. Based on these observations, we propose learn utilize from rich literature field classification further development As demonstration, group features, including local binary patterns, Markov neighborhoods cliques, Laws' masks, have selected form new set 22,153 which are used with FLD-based ensemble classifier steganalyze HUGO BOSSbase 0.92. At embedding rate 0.4 bpp (bit per pixel) an average detection accuracy 83.92% achieved. expected this approach can enhance our capability