An Image Compression Method of Fractal Based on GSOFM Network

In this paper, we proposed a method which incorporated multi-scale analysis into neural nets to solve the problem that fractal coding allows fast decoding but suffers from long encoding times. This method can reduce the computational load of fractal image coding significantly though efficient classification of image improve speed of image scan. Furthermore this paper also incorporates gray relational pattern analysis into the self-organizing feature maps (SOFM) network to develop a GSOFM network. The self-organizing feature maps network incorporated by gray relational pattern analysis is more effective and feasible than general method. In the appendix, we put forward the program of fractal coding, and carried out an artificial experiment. Experimental results and analysis show that proposed method can greatly speed up image coding process with a little worse image quality and compression ratio compared with the exhaustive search method. And it is better than Fisherpsilas method in image quality, compression speed and compression ratio.