Performance Analysis of Image Content Identification on Perceptual Hashing

Perceptual hashing can distinguish images of different contents effectively,and it is commonly used in content identification for detecting piracy or duplicate images.Published research of image content identification on perceptual hashing has mainly focused on the design of perceptual hashing algorithm,and very few theoretical methods have been developed for performance evaluation of content identification.Based on anti-collision property of image perceptual hashing,this paper introduces a mathematical model of content identification,defines performance formulas of content identification,and presents functions of the statistical distribution for performance evaluation.Experimental results show that the proposed approach estimates the performance of content identification on perceptual hashing very well.