Image Compression Research Based On Fractal Theory

Image has become one of the most important information carriers in the exchangesin people’s lives. The digital image must be efficient compression in order to meet thedigital image applications in various fields. Therefore, research on the efficient imagecompression technology is of positive practical significance. Recently, as more andmore new theories, new methods continue to emerge, resulting in a large number of newimage compression coding method. Based on fractal image coding method has a uniqueidea, a higher compression ratio, has a huge potential and other advantages. It is apromising compression method. But the current algorithm for fractal imagecompression has a problem that compression speed is very slow, it is difficult to meetthe needs of practical applications.In this paper, on the basis of previous research, theory and analysis of fractal imagecoding for digital image, a parametric search method based on the characteristics ofmathematical statistics-Key value fast search algorithm was proposed. With thefeatures of invariance of Key value before and after the affine transform of a matrix, thealgorithm has been improved the speed of fractal coding greatly. Experiments show that,on the premise of without reducing the quality of the reconstructed image, Key fastsearch algorithm which based on image statistical characteristics greatly accelerate theencoding speed. The Key value search algorithm reduce the search time, at the sametime, it can still guarantee search to match the block is the global optimal. It solve aproblem that many previously proposed algorithms cannot be both get the optimation insearch time and the best matching block.In addition, this paper to do a lot of research which the field used to do littleresearch, such as based on the fractal coding transform coefficients optimization, fractalcoding combined with embedded zero-tree wavelet coding and so on. Transformcoefficients optimization can make the fractal block coefficient gathered an optimal area.It can effectively improve the follow-up entropy coding compression ratio. The mode ofthe combination of fractal coding and embedded zero-tree wavelet coding can be maketwo algorithms complementarity, so as to achieve the1plus1is greater than2results.Experimental results show that, with the same conditions the proposed two algorithmscan significantly improve the compression rate. In this paper, all the work have beendone for the fractal coding, made a positive contribution to the development of fractal coding.