摘要
We propose a fragile watermarking scheme capable of image tamper wise dependency mechanism. Initially, the image is divided into detection and recovery with a block blocks with size improve image tamper localization precision. By combining image local properties of 2~2 in order to 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.
We propose a fragile watermarking scheme capable of image tamper wise dependency mechanism. Initially, the image is divided into detection and recovery with a block blocks with size improve image tamper localization precision. By combining image local properties of 2~2 in order to 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.
