Blind detection of malicious alterations on still images using robust watermarks

Digital image manipulation software is now readily available on personal computers. It is therefore very simple to tamper with any image and make it available to others. Insuring digital image integrity becomes a major issue. In this paper, we propose an original method to protect image authenticity using an invisible and robust watermark. Our scheme is independent of the signer, however the latter must have a high capacity and be able to extract the watermark in full blind detection mode. Our approach is based on the extraction of features from the image. These features are chosen so as to be unaffected by non malicious alterations such as lossy compression. They are embedded in the image using an iterative process so that watermarked image features and information contained in the watermark coincide perfectly. The authenticity is verified by comparing the features of the tested image, with those of the original image recovered from the watermark.