Fraud detection and self embedding

In this paper we address the issue of fraud detection in multimedia distribution. We use the self embedding approach to solve the problem. Self embedding serves to detect and recover possible tampered multimedia data. To use this approach, we design a powerful high capacity data embedding algorithm with l/167(1 bit out of 167 raw image bits)-l/84 hiding ratios subject to JPEG compression with the quality factor equals to 75. With such high embedding capacity, one can easily embed some important regions of an image unto the image self. We will demonstrate how the tampered data are recovered by using this technique.