Approximate regeneration of image using fragile watermarking for tamper detection and recovery in real time

Digital Image Watermarking has been rigorously used for image authenticity problems. Fragile watermarking is sensitive to even slight modifications and can detect tampering. This paper proposes a real-time tamper detection algorithm and regenerates an approximate image for the tampered locations. The image is divided into 4 parts and pair-wise correspondence is formed. This correspondence is used for marking the authenticity of parts of images and helps in the reconstruction of tampered parts. For every part, sub-blocks of size 4 x 4 are formed. Upon calculating the average of 4 inner pixels of these 4 x 4 blocks, a Local Binary Pattern (LBP) is constructed for that sub-block. Using XOR operation on 12-bit LBP, a 6-bit watermark for that sub-block is generated. This watermark is used for block authentication. Next, for recovery information, the 4 x 4 sub-block is divided into four 2 x 2 parts, and their average is calculated. This information is used for the recovery of tampered regions in the image. The watermark along with the average is then embedded into the corresponding sub-block in the pair-wise section of the image. This helps ensure the authenticity of the sub-block and helps in storing information needed for regeneration in case a part of the watermarked image has been tampered with. The imperceptibility and extent of degradation of the original image upon embedding the watermarking bits are calculated using the peak signal-to-noise ratio and structural similarity index. The approach is tested against a variety of attacks and shows its applicability in real-time applications with quick approximate reconstruction.