Distributed Fusion Steganalysis Based on Combination System Likelihood Function

This paper presents a distributed fusion decision-making steganalysis method for attacking image steganography. The fusion rule of combination system likelihood function is derived based on the smallest error rate Bayesian decision-making rule. According to information fusion theory, by introducing fusion decision-making theory into steganalysis research, it will achieve more effective, more reliable and more stable steganalysis detection effect than just adopting a single local decision-making. The image characters are first extracted from spatial domain, DCT domain and DWT domain, and then the final result is obtained by using fusion rule. The simulation results show the fusion steganalysis method significantly improves the detection performance of local decision-making through introducing the fusion rule of combination system likelihood function.