A reversible image watermarking algorithm for tamper detection based on SIFT

The integrity of the digital image content is very important for special types of images. To better locate tampered regions in the image, this paper proposes a reversible watermarking technology for tamper localization based on SIFT. By dividing the original image into blocks, the feature information of each sub-block is computed as the authentication watermark. The carrier image is then processed to extract the feature region of the image, and the authentication watermark generated by each sub-block is embedded into the corresponding feature region through mapping. When the watermarked image is attacked, the related attack is detected by the invariant moment, and the sub-block feature value and the authentication watermark extracted from the feature region are calculated to determine if the sub-block has been tampered with, and if so, the position of the tampered region is given. The experimental result illustrates this scheme can detect and locate the tamper region effectively and accurately, and the scheme has certain robustness and invisibility.