一种新的可实现图像篡改检测和恢复的脆弱水印算法
摘要
We propose a fragile watermarking scheme capable of image tamper detection and recovery with a blockwise dependency mechanism. Initially, the image is divided into blocks with size of 22 in order to improve image tamper localization precision. By combining image local properties with human visual system, authentication data are acquired. By computing the class membership degree of each image block property, data are generated by applying k-mean clustering technique to cluster all image blocks. The recovery data are composed of average intensity obtained by truncating the two least significant bits (LSBs) of each pixel within each block. Finally, the logistic chaotic encrypted feature watermark consisting of 2-bit authentication data and 6-bit recovery data of image block is embedded into the two LSBs of each pixel within its corresponding mapping block. Experimental results show that the proposed algorithm does not only achieve superior tamper detection and locate tiny tampered positions in images accurately, it also recovers tampered regions effectively.