如何快速有效地为高频复杂区域找到合适匹配块是分形图像压缩中一个难以解决的问题。考虑到遗传算法的可并行性与全局搜索能力,结合匹配搜索的特点与要求,提出了一种逐位变异、最优保留的紧凑遗传算法,并将这种方法应用到高频复杂区域的匹配搜索中去。分析表明,这种算法具有较之其他随机搜索算法更好的收敛性、更高的搜索速度和全局搜索能力,能够大大提高匹配搜索中最优匹配块的捕获能力和搜索效率。实验结果也证明了这种算法在分形图像压缩匹配搜索算法中应用的优越性。

How to find the optimal match block for a complex region block with high frequency signals more quickly and efficiently is a very difficult problem in fractal image compression.Considering of the parallel characteristic and the global seeking ability of genetic algorithm,combining with the characteristics of and the requests for match seeking,this paper proposes a compact genetic algorithm mutated by bit with holding the optimal individuals and applies the algorithm in the match seeking of high frequency regions.Analysis of the algorithm shows that this algorithm has much higher convergence ability,seeking speed,and global seeking ability than other random seeking algorithms.It can greatly improve the obtain ratio of the optimal match block and the seeking efficiency.Experimental results also show the superiority of its application in match seeking of fractal image compression.