An HEVC steganalytic approach against motion vector modification using local optimality in candidate list

Abstract Current research of video steganalysis is mainly oriented to H.264 or even earlier formats. As the H.265/HEVC standard is becoming more and more popular, there have been steganographic methods specifically designed for HEVC videos. Previous steganalytic methods are no longer effective due to the new features of HEVC. The advanced motion vector prediction technique employed by HEVC provides a new way for motion vector (MV) modification, utilizing the index of the candidate MV list. Such modification cannot be detected by previous steganalytic methods because the value of MV is not changed. To solve this problem, we proposed a new steganalytic strategy employing not only the local optimality of MV but also the local optimality in the candidate list. Combining the two types of local optimality, a 40-dimensional feature set is designed. Extensive experiments are carried out to evaluate the effectiveness of the proposed feature set. It is shown that compared with three current steganalyzers, our approach improves the performance against typical MV modification methods under various settings. In particular, our approach significantly boosts the detection accuracy of the index modification method that is exclusive to HEVC videos. To our best knowledge, this is the first work of video steganalytic method exploiting the new features of HEVC format.