A Steganalytic Approach to Detect Intra Prediction Mode Modification Using Difference of Partitioning Structure for HEVC

Consumer electronics offer significant convenience for online communication and entertainment. However, Consumer electronics also encounter data security issues, as the data can be easily used as steganography cover for convert communication, particularly the video data, which provides large capacity and is difficult to detect. To address this issue, a steganalytic approach using the difference of partitioning structure is proposed to detect Intra Prediction Mode (IPM) modification for H.265. We find that many optimal intra prediction modes are modified to sub-optimal intra prediction modes during the procedure of IPM-based steganography, which makes the partitioning structure tends to be unstable after calibration. Meanwhile the partitioning structure in I frame tends to be more stable after calibration if no message is embedded. In order to detect IPM modification, Firstly, the video sequence is calibrated. After that, all the partitioning structure and IPM information can be obtained in both original video and the calibrated video. Then, depth matrix of each Coding Tree Unit (CTU) can be calculated. Finally, 11-dimensional feature sets are extracted based on the difference of partitioning structure. Experimental results indicate that the proposed feature sets have good performance against IPM-based steganography for H.265.