Multiple Image Watermarking in DWT Domain Using Neural Network

This paper proposes the formulation of a more secure image watermarking methodology based on exploitation of the perpetual capacity of Discrete Wavelet Transforms (DWT) and reversible compression capability of Counter Propagation Neural network (CPN). The usefulness of CPN in mapping watermark images into few bits and subsequent embedding with the cover images in wavelet domain is demonstrated. The proposed method exhibits high security,high capacity and high imperceptibility features. Low embedding and tracking costs and low probability of coincidence are making this method unique among the previously investigated techniques by various researchers.