Reversible authentication and cross-recovery of images using (t, n)-threshold and modified-RCM watermarking

This paper presents an image authentication scheme for digital images. The proposed scheme protects a group of n images mutually. It designs a block matching procedure to generate the recovery data for each image, and applies a (t, n − 1), 2 ≤ t < n, threshold mechanism to encode each recovery data in n − 1 shares. The recovery shares are cross-embedded in the n images using a modified reversible contrast mapping watermarking scheme. A signature-based authentication code is finally generated and stamped to provide evidence for integrity of each image. The scheme not only can detect the tampering activities, but also can locate and recover the invalid regions of the tampered image if t or more watermarked images in the same group were intact. A nice characteristic of the proposed scheme is that the original images can be reconstructed lossless if no watermarked image was tampered, making the technique feasible in the application of protecting very sensitive images such as military or medical images. Experimental results show that the proposed scheme successfully detects various kinds of image alterations such as filtering, cropping, and replacement, and the corrupted images are properly recovered using the cross-recovery scheme.