A New Video Steganalysis Algorithm against Motion Vector Steganography

In contrast to steganography, steganalysis is the art of detecting the presence of the hidden information transmitted through the cover object. It has currently become a hot spot in the field of information security. In this paper, a video steganalysis technique is proposed against the video stegnanography methods which hide the secret message by a least modification of the motion vectors in the compressed bitstream. The method utilizes the feature classification technique, based on the analysis of some statistical properties, to determine the existence of the hidden message, in which the support vector machine (SVM) is then used as discriminator. Experimental results show that the proposed method can effectively detect the embedded data introduced by motion vector steganography algorithms.