Digital Image Watermarking Based On Fractal Coding

Fractal, called by the "Nature's geometry", has a huge advantage in describing the nature of its various forms. In 1990s, fractal theory was applied to image processing. Stemming from fractal image compression, fractal coding has penetrated into digital watermarking, feature extraction, face recognition, database retrieval, image denoising and other areas.This thesis describes the theoretical foundation of fractal coding, and analyzes the classical and orthogonalization fractal coding algorithm, and then studies the characteristics of their code. Experimental result shows the replaceable property of the fractal optimal matching blocks (OMBs), i.e. changing the OMBs whose collage errors are small will not affect the image quality. Both theoretical analysis and experiments prove that orthogonalization fractal coding's luminance offset is invariant. Meanwhile the image is continuously dependent on the luminance offset. By experiments, we find the principle of the differences between decoded image coding and original coding.According to the code's features, this thesis presents following three new digital watermarking algorithms. All these algorithms need to do fractal coding on the host image and the watermark. And we embed the code of the watermark into the host image's fractal code.Methodl:The host image's OMBs is replaced by the watermark's OMBs di-rectly. Then we change the host image's contrast scaling and luminance offset. And the contrast scaling and luminance offset of the watermark are preserved as a key. The detection process needs the key and the original host image. Method: We apply methodl to orthogonalization fractal coding, and embed the key into the host image's luminance offset. The detection process only needs the original image. Method3:The first step we also apply methodl to orthogonalization fractal coding, then we embed the key and location information into the host image. The detection process needs nothing but the image to be detected.