Robust Hashing Based Video Content Authentication

摘要

Nowadays, robust hashing has been widely adopted in video content authentication. The hash is resistant to common video processing operations, but vulnerable to video malicious attacks. Existing hashing methods to authenticate video, however, usually fail to make a good trade-off between robustness and security. In this work, a novel robust video hashing based on three-dimensional gravity centroid (3DGC-based hashing) is proposed for video authentication. Video frames are firstly normalized and partitioned on the basis of Hausdorff distance. Sub video cubes are generated from grouped video frames in both space and time domain. Then centroid of three-dimensional gradient are extracted from these cubs and quantized to form the video hash. During the experiments, parameters settings of the 3DGC-based hashing are discussed and the results show that the proposed method is superior to other methods in terms of security and robustness.