摘要
分析了应用于多峰值多欺骗性适应度函数的快速的遗传算法问题,提出基于性能优良的最优家族遗传算法(OFGA)的改进方法,使其个体的进化仅仅是基于优良家族而不是全部种群以克服早熟。最后,将改进后的算法应用到快速图像匹配以证明有效性,在图像匹配的适应度函数构造中,为减少非匹配点计算量,提出采用序贯相似性检测算法(SSDA),实验结果表明,改造过的OFGA和SSDA相辅相成、互相受益,整体算法对提高图像匹配的速度方面成效显著,且算法稳定,说明算法有应用到类似问题上的潜力。
Based on the analysis of the speed and stability of the genetic algorithm applied to functions with multi-modality and multi-deceptive-problem, the improvement on powerful genetic algorithm （family genetic algorithm） is put forward that individual evolvement is just based on not the whole population but the optimal family to avoid the premature phenomenon. At the same time, the new algorithm is applied to image matching to prove the improvement effective. In order to reduce the calculation amount on non-optimum points the sequence similar detection algorithm （SSDA） is introduced to be the fitness function. The experimental results indicate that improved optimum family genetic algorithm and SSDA can be benefited from each other. The whole algorithm is great effective in improving the speed of image matching and its performance is steady. It can conclude that the new algorithm is potential in solving the similar problems.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2005年第10期1027-1030,共4页
Chinese Journal of Scientific Instrument
关键词
图像匹配
遗传算法
适应度函数
Image matching
Genetic algorithm
Fitness function
