Robust watermarking through feature-based watermark synchronization

Abstract

This thesis addresses a novel robust watermarking method for digital images by using feature-based watermark synchronization. Most previous watermarking algorithms are unable to resist geometric distortions that desynchronize the location where copyright information is inserted. We propose two watermarking schemes that are robust to geometric distortions. In order to resist geometric distortions, we use local invariant properties of the image by applying (1) the circular Hough transform (CHT) and (2) the scale invariant feature transform (SIFT). These transforms are invariant to translation and scaling distortions. The watermark is inserted into the circular patches generated by (1) CHT and (2) SIFT. Rotation invariance is solved using the translation property of the polar-mapped circular patches. Our schemes belong to the blind watermark, because we do not need the original image during watermark detection. We have performed an intensive simulation to show the robustness of the proposed schemes. The simulation results support the contention that our schemes are robust against geometric distortion attacks as well as signal processing attacks listed in StirMark 3.1. We have compared our results to other methods, which comparison has shown that our schemes outperform the previously proposed methods.