Research On Fractal Image Compression Methods

Nonlinear science is a foundational discipline which concerns the common properties of nonlinear phenomena. And fractal theory is one of important subdisciplines of nonlinear science. The research studies several methods to improve fractal image compression.A texture correlation and the ICA search strategy are utilized to speed up the encoding process and improve the bit rate for fractal image compression. Texture feature is one of the most important properties for the representation of an image. Entropy and maximum entry from co-occurrence matrices are used for representing texture feature in the image. For a range block, concerned domain blocks of neighboring range blocks with similar texture feature can be searched. In addition, domain blocks with similar texture feature are searched in the ICA search process, which further makes use of the texture correlation between blocks and speed up the encoding process.A fast and efficient no-search fractal image coding method based on a modified gray-level transform which uses a fitting plane is presented. The improved gray-level transform can reduce the minimum matching error between a given range block and its corresponding domain block, and thus, it can enhance the possibility of successful domain-range matching. In comparison with the gray-level transform which uses an adaptive plane, the improved scheme results in a considerable acceleration of the encoding process, decreases the compression ratio and improves the quality of the reconstructed images in the meanwhile. Comparing with Furao's no-search scheme, our improved scheme can get higher PSNR at higher bpp, and it achieves almost the same PSNR at lower bpp. Although it uses more transform coefficients, such a fitting plane method can speed up the encoding process with the quality of the reconstructed images improved.A novel no search fractal image coding approach based on conversion is proposed in this paper. We do not encode the original image directly but convert it to two special images, then encode them using no search fractal image method based on two gray-level transforms respectively, one for the large blocks and the other for the small blocks based on the quadtree partition scheme, meanwhile a precondition is introduced to see whether the range block is a constant block. We analyze and prove that it reduces the minimum conversion matching error of the range block in theory. In the end we convert the two decoding images to one. Experiments on standard test images show that the improved scheme speeds up the encoding time, improves the compression ratio and enhances the quality of the reconstructed image. Even if only dealing with the first image got from the original image, it also works well. Further more, the improved scheme can maintain a good balance for general images.