Color Wavelet Based Universal Blind Steganalysis

This paper presents a universal blind steganalysis method for color images. The key new components of the presented method are color wavelet decomposition and a set of feature vectors. In addition, it is based on simple Fisher Linear Discriminatant approach for classifying the received image as containing steganographic content or not. The presented class of color wavelets has several useful properties and has been shown to be effective at detecting stego data within digital color images. Testing and analysis was conducted with 200 color images varying in size and style. Experimental results have shown that the performance of the presented method is better than some popular steganalysis methods.