An Improved Steganalysis Method of LSB Matching

In this paper, we propose an improved steganalysis algorithm to detect spatial domain least significant bit (LSB) matching steganography, which is much harder than the detection of LSB replacement. Firstly we propose run length based features, which are sensitive to LSB matching in spatial domain of images. At the same time, we extend Kerpsilas features, the statistical moments, to higher orders, which are sensitive to LSB matching in transform domain. Then the improved method is constructed based on these features. Experimental results on two datasets demonstrate that this method has superior results compared with other recently proposed algorithms, and shows that the proposed method is efficient to detect the LSB matching steganography on compressed or uncompressed images.