An Image Structure Information Based Robust Hash for Tamper Detection and Localization

Abstract

Although image robust hashing has drawn many attentions as a promising image authentication technique, the tamper detection and localization of robust hash algorithm has not been explored thoroughly. The most important difficulty is the conflict that the hash value is supposed to be robust to common content-preserving manipulations (CPM), such as lossy compression, but sensitive to malicious tamper. To overcome the contradictory requirements, a general robust hash based image tamper detection and localization scheme using existed robust hash algorithms is presented, considering the different features of common CPM and malicious tamper. Under the scheme, a new structural information based image robust hash algorithm is also presented. The experimental results over a public image database show that our algorithm outperforms some representative algorithms. Different from previous literature, the tamper detection and localization still works well under some typical CPM.