Image authentication and tampering localization based on watermark embedding in the wavelet domain

In this study, an image authentication and tampering localization technique, which is based on embedding a digital watermark into a plain image in the wavelet domain, is proposed. First, a previous blind watermarking technique is adopted to be robust to image compression. Then the extracted watermark is used to determine whether the watermarked image has been modified or not. If the image has been modified, the proposed sequential watermark alignment (SWA) and the morphological region growing (MRG) methods are used to find the tamper parts and determine the tampered region, respectively. Simulation results verify that several kinds of tampering effects in images can be correctly and efficiently detected by the proposed methods.