RST Robust Watermarking Schema Based On Image Normalization and DCT Decomposition

Abstract

Geometric distortions are generally simple and effective attack to many existing watermarking methods which can make detection of the embedded watermark difficult or even impossible. A robust watermarking system must be able to encounter such attacks based on rotation, scaling and translation operators (RST attacks). In this paper, we describe a new robust watermarking schema based on logo embedding in the DCT transformed domain using image normalization techniques. In contrast to existing approaches, the watermark is not embedded directly in the normalized image. The image normalization is just used for calculating the affine transform parameters so that the watermark embedding and detection is performed in the original coordinates system. The performed experiments show that the proposed algorithm is robust against various types of attacks such as low-pass, median, Gaussian noise, aspect ratio change, rotation, scaling, JPEG compression, and the combination of these attacks.