Digital watermarks offer means of protecting copyright of digital multimedia. However, many of the proposed watermarking methods are vulnerable to the geometric distortions which occur during normal use of the media. For that reason, watermarking methods resistant to geometric distortions are required. In this thesis, we review this problem and propose two watermarking schemes based on invariant pattern recognition theory.
We propose an invariant watermark using the Radon transform and higher order spectra. A bispectrum feature vector of the image is used as the watermark. Our approach differs from the previous methods in that we embed watermark into the phase of the higher order spectra. Also, our Radon embedding grid outperforms the Fourier-Mellin based methods. We devised a new embedding method which allows detection of the watermark when there is no exact inverse function during embedding. As we use the Radon transform, our method can be used for medical images. We show the invariance of the designed watermark with mathematical proofs. Experimental results confirm that this scheme is resistant to geometric distortions.
We also propose an invariant watermark using the Zernike moment. Zernike moment is an algebraic invariant and used in various applications in pattern recognition and image processing. The rotation, scale and translation invariance is achieved by the normalized Zernike moment of an image. The watermark signal is embedded into the Zernike moment of the image. Our method is more robust than the other moment-based methods as the Zernike moment has better resilience to noises. We show the invariance of the watermark using mathematical proofs. Experimental results confirm that this scheme is resistant to geometric distortions.
Finally, we present the benchmark test results of the proposed methods. StirMark3.1 is used for the benchmark test. Comparative results show our methods outperform the other commercially available schemes.