Detection of fake high definition for HEVC videos based on prediction mode feature

摘要

A high definition (HD) video usually means a good visual quality, and the video with low quantization parameter (QP) or high bit rate has a good video definition. However, forgers prefer to directly re-encode lower definition videos using lower QP or higher bit rate without video quality improvement to pretend to be HD videos. Therefore, fake HD video detection is necessary in video forensics. In this paper, a novel method is proposed to detect the fake HD for high efficient video coding (HEVC) videos based on prediction mode feature (PMF). According to our analysis, the HEVC prediction mode scheme of the subsequent encoding could be influenced by the previous lower quality encoding, hence the frequencies of prediction unit with all types of prediction modes can be used to detect the fake HD videos. Firstly, a 4-D features are extracted from Planar, DC, H0 direction and V0 direction intra prediction modes. Secondly, a 6-D features are extracted from Skip, Merge and AMVP inter prediction modes. Finally, these two feature sets are combined into the PMF to detect fake HD videos and further estimate their original QPs and bit rates. Experimental results show that the performance of the proposed method outperforms state-of-the-art works. (C) 2019 Elsevier B.V. All rights reserved.