Image authentication using perceptual hashing

摘要

This paper presents a review of image authentication methods, based on the perceptual image hashing approach. Firstly we analyze two algorithms, which have the capability of determining if an image is authentic or not, even if it has suffered content preserving distortions such as compression, filtering and other signal processing operations; as well as some malicious modification such as geometric distortions. Next two modifications of these algorithms are analyzed which improve their performance by increasing their robustness against geometric distortions providing them also tamper detection capability. Finally two recently proposed algorithms with not only the tamper detection capability, but also with the capability of localizing tampered regions are described. Evaluation results are given to show the performance of these methods under analysis.