ISC : Steganographer Detection Based on Internal Spectral Clustering

Steganographer detection aims to find guilty users who hide secret information in images on social work. In existing work, the distance between users is calculated according to the distribution of all features extracted from corresponding users, then the user who deviates most from others is defined as the guilty user. The challenge is that when the embedding rate of stego images is low, the difference between users is reduced, and it is difficult to locate the guilty users. Aiming at the distribution difference between stego images and cover images within users, a steganographer detection algorithm based on internal spectral clustering (ISC) is proposed in this paper. ISC uses graph-based spectral clustering method to divide the user’s spread image into two clusters, and then feeds the result into the discriminant function to locate the guilty users. The simulation results illustrate that the proposed ISC method is superior to the state-of-the-art method both in effectiveness and efficiency.