A Multi-Channel Combination Method of Image Perceptual Hashing

Traditionally, the proposed perceptual hashing algorithms are mainly focused on the feature extraction and coding stages. But they only concerned parts of perception information, and take a certain type of features to obtain codes, which makes them to be biased on resisting some varieties of attacks. On the other hand, HVS (Human Visual System) as one of best methods simulating the image cognitive processing of human vision system normally takes into account more than one perception factor. Based on the human nature of image cognitive processing and the conclusion of HVS, we suggest that image perceptual hashing should take the advantages of different perceptual feature channels and reserve the superiorities of each. In this paper, a multi-channel combination method is proposed and the performance under many general attacks is examined.