Robust Image Hash Based on Cyclic Coding the Distributed Features

Abstract

Robust hash measure the similarity between two images by matching their short hash vectors. For its advantages of efficiency and versatility, robust hash has become a promising technology in image authentication and identification. Different to the traditional watermarking method, robust hash does not need to embed information into the host image. Therefore, it is promising to provide more robustness to geometrical distortions by robust hash. In this paper, a novel robust image hash algorithm is addressed on the issue of image rotation, which coding the distributed image features in cyclic manner, and the performance is discussed by an illustrative experiment.