Binary Image Steganalysis Based on Local Residual Patterns

Abstract

A binary image steganalysis scheme based on the statistic model of local residual pattern (LRP) is proposed in this paper. LRP means the pattern of a local area of the residual map of the binary image, which is calculated with the XOR operation. The XOR operation is sensitive to the difference between adjacent pixels, which leads to the emphasis on edge property of the residual map. The neighbouring LRPs of the modified pixel will be affect, which makes the statistic model of LRPs change. Thus the trace of steganography can be detected according to the difference between the statistic models. Finally, the experiments we conducted show that our proposed scheme is effective on binary image steganalysis.