Self-embedding framework for tamper detection and restoration of color images

The present era is paving huge expansion to the transmission of digital data in fields like health, military intelligence, scientific research, and publication media, etc. The nature of digital data makes it more vulnerable to various intentional and unintentional attacks and hence increases the necessity of verifying its integrity and authenticity. In this paper, a novel image watermarking method for tamper detection and recovery of color images has been proposed. The color image is primarily separated into three planes, and then each plane is divided into four equal halves. We sub-divide each half into 4 x 4 non-overlapping blocks. Further, from a group of four corresponding sub-blocks, a 32-bit watermark comprising of the arithmetic mean value along with the 8-bit data containing the location of the mapped block is generated. This 32-bit generated watermark is embedded into the two least significant bits (LSB) of the mapped block after being encrypted using gray code. The information is embedded in such a way using chaotic sequence such that a full restoration process can be carried out, even if 75% of blocks in any of the three planes get tampered with. This method obtains a high-quality restored image with an average PSNR value of 39.22 dB for the RGB model and 39.37 dB for the YCbCr model. The performance of the proposed method for various collage attacks shows its efficacy over other state-of-art techniques, making it a suitable candidate for the recovery of tampered color images.