Image Restoration-Based Steganalysis Directed to LSB Matching Steganography

The least significant bit(LSB) matching steganography is one of the most difficult and attractive subjects to steganalysts.To achieve better detection performance,this paper proposes a steganalysis algorithm applied to spatial image,by combing the image degradation/restoration theory with steganalysis.By the algorithm,firstly we model the image degradation process caused by additive noise,on embedding process,and proposes a spatial restoration filter correspondingly.In case the restored image of stego-image is taken as the estimation of the cover-image,we calculate the center-of-mass feature vector between the original image and the restored one.And the steganalysis is accomplished by introducing the mentioned feature vector into Fisher linear discrimination.Experimental results show that the proposed filter manages to recover the stego-degraded image,and the steganalysis method generally performs better than Ker's method,especially in the condition of low embedding rate.The proposed steganalysis algorithm is applicable to LSB matching steganography for spatial images.