Abstract: The well-known adage that "seeing is believing" is no longer true due to the pervasive and powerful multimedia manipulation tools. To prevent image manipulations and fraudulent use of modified images, multimedia authentication techniques are being developed. But nfortunately there is a big problem in authentication watermarking algorithms which are in the transform(e.g., DCT) domain. Converting the modified transform coefficients back to the integer pixel domain and then converting them again to the transform domain may not have been manipulated otherwise. Combining the embedding process with the performance analysis, a new semi-fragile authentication watermarking with feedback, called SAF, is presented. The embedding process in some image blocks is performed several times according to the results of the performance analysis of the watermarking detecting process. By the reconstruction of the DCT coefficient, SAF can accept JPEG lossy compression on the watermarked image to a pre-determined quality factor, and reject malicious attacks. The experimental results show its effectiveness in protecting the content-integrity of images.

Key words: authentication watermark, semi-fragile watermark, watermarking algorithm