A new approach to reliable detection of LSB steganography in natural images

In this paper, a new steganalysis technique is proposed on the basis of statistical observations on difference image histograms for the reliable detection of least significant bit (LSB) steganography. A physical quantity is derived from the transition coefficients between difference image histograms of an image and its processed version produced by setting all bits in the LSB plane to zero. It appears that this quantity is a good measure of the weak correlation between successive bit planes and can be used to discriminate stego-images from cover images. Further studies indicate that there exists a functional relationship between this quantity and the embedded message length. Based on those facts, an equation is formed to estimate the amount of embedded messages. Experimental results show that the proposed algorithm is comparable to previously proposed techniques. The new approach is applicable for the detection of both random and sequential LSB embedding. (C) 2003 Published by Elsevier B.V.