A new semi-fragile image watermarking with robust tampering restoration using irregular sampling

摘要

This paper presents a semi-fragile watermarking method for the automatic authentication and restoration of the content of digital images. Semi-fragile watermarks are embedded into the original image, which reflect local malicious tampering on the image. When tampered blocks are detected, the restoration problem is formulated as an irregular sampling problem. These blocks are then reconstructed, making use of the information embedded in the same watermarked image, through iterative projections onto convex sets. In contrast to previous methods, the restoration process is robust to common image processing operations such as lossy transcoding and image filtering. Simulation results showed that the scheme keeps the probability of false alarm to a minimum while maintaining the data integrity of the restored images.