JPEG steganalysis based on the entropy features of wavelet histogram distributions
To detect the existence of hidden message in JPEG images, the histogram distributions of wavelet subbands are captured to describe the images statistical characteristics. And the symmetrical cross entropy difference features of the histogram distributions that captured from both detection images and their estimated cover images are calculated. Then a feature-matched one-class classifier is designed to make classification. Experimental results demonstrate that the proposed method can effectively distinguish between cover images and stego images, and significantly improves the detection accuracy.