An Improved Approach in PSO-based Fractal Image Compression

Abstract

An approach is proposed in this paper to achieve fractal image compression based on the fractal quadtree, particle swarm optimization (PSO), and Gaussian function by using fitness function. Since the encoding speed of the traditional full search method is time-consuming, the fast evolutionary PSO approach based on Gaussian function is proposed to speed up the encoder and preserve the image quality. The PSO approach uses the properties of fractal quadtree to reduce the time-consuming full search method. In PSO, every block of the divided image is regarded as particle and to be directed to the region which is consisted of a candidate of higher similarity. Finally, the particles are replaced by the region and the purpose of fast image compression can be achieved.