Predigest Watson's Visual Model as Perceptual Hashing Method

Abstract

HVS (human visual system) has been widely used to evaluate image quality over the past three decades, which provided a good simulation on the image cognitive processing of human vision system. As a newly developing technology, perceptual hashing need to take the perception information into account during the feature extraction stage, and then code them into a short digest. In this paper, we present a new image perceptual hashing method by predigesting the well-known Watson's visual model. By comparing with the Watson's visual model and other perceptual hashing method, our new method takes sufficient perception information into account and has better performance of robustness and discriminability.