Medical image authentication based on fragile watermarking using hamming code

摘要

In this paper we propose a novel fragile watermarking scheme that uses very little bit budget to detect tamper in medical images, based on hamming codes. Unlike other papers that used hamming codes in a robust watermarking context, we use this code to detect errors in received watermarked image. The host image is decomposed into blocks and verification is performed block-wise in a way that we can easily localize tampers. A pair of keys is used for security reasons and watermark bits are placed into the blocks in accordance with these keys. Experimental results show that this method can reach near perfect verification while having very high PSNR and thus making very little changes to the host image that is of crucial importance in medical images.