Watermarking using complex wavelets with resistance to geometric distortion

We present a watermarking algorithm in the complex wavelet domain and show why complex wavelets are better than real wavelets. Being an oversampled transform, the Complex Wavelet Transform requires special precautions during watermark embedding. We then model the watermarking process as a communcation channel and our results show that the complex wavelets domain has relatively higher capacity than both the spatial and the real wavelets domains. We will also outline a motion-based algorithm for image registration, which can help recovering watermarks from images suffering from geometric distortion.