多家族遗传算法及其在图像匹配中的应用  

Mul-families Genetic Algorithm and the Application for Image Matching

作　　者：张金萍[1] 倪洪启[1] 王立强[1] 于玲[1] 

机构地区：[1]沈阳化工学院机械工程学院

出　　处：《沈阳化工学院学报》2008年第1期77-80,共4页Journal of Shenyang Institute of Chemical Technolgy

摘　　要：提出一种改进的遗传算法——多家族遗传算法.算法采用实数编码方式,每一代由若干家族构成,通过家族之间的相互竞争、选择和交换与各家族内部的交叉选择完成遗传操作,并引入不对称交叉方式,该方式使每一代的搜索范围为整个参数空间,并免除了独立的变异操作.算法在4个典型寻优问题的应用中表现出很强的抗早熟能力和较快的收敛速度,并成功地应用于图像的匹配寻优当中,匹配速度和匹配精度均较为理想.A novel, multi-families genetic algorithm, is presented. The real coding is used in this algorithm and several families compose every colony of algorithm. The process of algorithm comprises of competition,selection and exchanging between every family and the crossover and selection in every family. At the same time, the anisomerous crossover is used to make the search space is whole space. The multifamilies genetic algorithm is used to resolve four classic problems, the results show that the novel algorithm＇s convergence rate,success rate and anti-premature all have evidence superiority. At last, the novel algorithm is applied for image matching, and the precision and rate of matching is excellent.

关 键 词：遗传算法 寻优 收敛 图像匹配 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]