Detection Of Double Compression Of HEVC Videos Based On PU Block Partition Modes And Co-occurrence Matrix Of DCT Coefficient

Due to the popularization of network applications and the rapid development of multimedia information industry,digital videos get access to people's life through the Internet and smart phones,and more importantly,they have become a vital part of judicial evidence.However,the integrity and authenticity of digital video has become an increasingly severe issue.Easy-to-use multimedia editing software with powerful function has rendered digital images and videos exposed to forgery,which influences the judicial justice and the social stability.Therefore,the identification of digital video authenticity will have essential practical value and bright future.To tamper with videos,they must experience double compression,which makes the video double compression forensics technology become an important research field in the detection of video authenticity.HEVC,as the latest video coding standard,will be widely used in high-definition,ultra-high-definition video coding.In this paper,two algorithms are proposed for the HEVC video double compression detection.1.The video double compression detection algorithm based on PU block partition mode is proposed to expose HEVC video re-compression.Based on the fact that the number changing of 4x4 PU block in HEVC double compression videos is inconsistent with corresponding single compression video,the number of 4x4 PU blocks of videos with single compression and double compression is analyzed and normalized.The curve of number change is extracted and the curve bulge can be used to detect double compression for HEVC videos.Experimental results show that the proposed algorithm can effectively detect HEVC doubly-compressed videos.This algorithm can extract the features of the PU syntax elements and achieve the detection of the HEVC re-compressed videos,which further promotes the development of video recompression forensics algorithm.2.The video double compression detection algorithm based on PU block partition mode and DCT coefficient co-occurrence matrix is proposed for double compression detection in HEVC videos.The quantized DCT coefficients of doubly-compressed HEVC video I-frames differ from the single compression video,which can be characterized through the co-occurrence matrix.The PU(Prediction Unit)partition mode and co-occurrence matrix of quantized DCT coefficient are combined to build a joint feature to reflect the data change comprehensively caused by double compression.Experimental results show that the proposed algorithm can effectively distinguish HEVC doubly-compressed videos from single compressed ones.