Parameter-estimation and algorithm-selection based United-Judgment for image steganalysis

In order to synthetically utilize multiple steganalytic algorithms, and further improve the detection accuracy and enhance detection reliability, United-Judgment methods are researched and analyzed in this paper. According to the performance of each algorithm, United-Judgment methods for both blind and specific steganalysis are proposed based on parameter-estimation and algorithm-selection. Experiments are carried out for the former with seven typical blind detections and the latter one with five typical spatial domain steganalytic methods. Experimental results show that the proposed methods can synthetically utilize the existing multiple algorithms effectively, and achieve more reliable detection.