A universal steganalysis to steganographic images on frequency domain

Current steganalysis for images is not universal, which is usually special for one kind of steganography and not practical for others. Meanwhile the correct detection rate of these detecting methods is not high. After analyzing current steganalysis methods, a new steganalysis method for detection of steganography in image´s frequency domain is proposed. The detecting method is universal and is not restrained to some special steganography method. Based on the spectrum analysis of difference histogram of frequency coefficients, the detection is achieved according to evident spectrum difference between non-steg-images and steg-images. The physical frame of the algorithm is easy and the detection method can be put into practice. Experimental results from detecting steganographic images of DCT domain and DWT domain show that the detection performance is satisfying. Compared to Stegdetect 0.6, which is a steganalysis system for JPEG images, the method shows much better performance.