Fractal-based techniques for a generalized image coding method

This paper presents a new generalized image block coding algorithm which covers fractal techniques, block transform techniques, and vector quantization as its special cases. The coding is performed by approximating each image block with a linear combination of a series of (not necessarily orthogonal) blocks selected from a pool of basis blocks. This pool is made up of (1) a set of fixed basis blocks, (2) a set of blocks taken from the filtered, subsampled image, and (3) a set of blocks taken from the image without any change of scale. The index of the selected basis blocks and their corresponding coefficients make the code for each range block. Methods are proposed for making the pool and selecting blocks from the pool.