Steganalysis of adaptive image steganography in multiple gray code bit-planes

For detecting the MBPIS (multiple bit-planes image steganography), which embeds the messages into the multiple Gray code bit-planes, a steganalysis method is proposed to estimate the embedding ratios based on sample pairs analysis. The proposed method combines appropriate trace sets of a more elegant and pellucid sample pair analysis model to estimate the modification ratio of each natural binary bit-plane. Then the modification ratios in Gray code bit-planes are estimated from the modification ratios in natural binary bit-planes based on the relationship between natural binary bit-planes and Gray code bit-planes. Finally, considering the matrix encoding in MBPIS, the embedding ratios are obtained from the modification ratios in Gray code bit-planes. Experimental results show that compared with existing steganalysis methods, the proposed method can not only detect the stego images of MBPIS more reliably, but also estimate the embedding ratios of MBPIS with smaller errors. When the embedding ratio in the first Gray code bit-plane is not less than 0.1, the average absolute error of the proposed method will less than it of the ID3SPA by 0.04 at least.