抄録
Fractal image compression is a unique image compression method based on self-similarity in images. The main disadvantage of this, however, is the high computational cost. Meanwhile, the particle swarm optimization (PSO) which is one of the techniques of biogeography based optimization attracts much attention in various computer science problems. In this study, we apply the PSO technique for the fractal image compression. Experimental results show that the PSO is about 70% faster than the traditional exhausted method. Furthermore, PSO algorithm is suitable for parallel computing, thus we also implement this program on Graphics Processors Unit (GPU) for more speedup. Finally the GPU-PSO mthod is about 60% faster than CPU-PSO method.
