Robust Image Hashing Based on Local Features for Image Authentication

Abstract

In this paper, an image hashing method based on local features is proposed. At first the input image is pre-processed and divided into un-overlapped blocks. We choose several blocks as the effective blocks using SIFT features. Then color, texture and shape features of the selected blocks are extracted, connected and permuted to form the final hash. Experimental results show that this method is robust against most content-preserving attacks. Collision probability of this method is smaller than other methods. This method can also be used to detect tampering image.