A Universal Steganalysis Method with Compressive Sensing in Image Spatial Domain

Based on the textural features of grayscale images, a scheme of compressive sensing universal steganalysis (CSUS) in spatial domain was proposed. Firstly, directional lifting wavelet transform (DLWT) was employed as a sparse representation, and corresponding sparse coefficient was used to calculate histograms of images. Then, measurement matrix of the compressive sensing(CS) was designed with the generalized Gaussian distribution (GGD) model, and the CS value was obtained by using the matrix to sense the sparse coefficients, which were regarded as the textural features. Finally, the classification of image steganalysis was implemented by the support vector machine (SVM). The steganography of four kinds of image databases were performed with five kinds of steganagraphic algorithms. Steganalysis was carried out with the proposed CSUS and classical steganalysis methods, and the results were analyzed and compared. Experimental results show that the proposed CSUS method is universal and has higher accuracy for detecting spatial domain steganography, and feature dimension can be reduced.