摘要
图像融合的评价指标多种多样,但往往不能同时达到最优,有时甚至是相互矛盾的,融合可以看作是多个评价指标最优化的过程。提出基于改进基因表达式编程的空域和多尺度域多目标优化图像融合算法,为解决图像融合多目标优化搜索空间太大的问题,提出多区域特征联合决策对种群进行初始化,改善初始种群的质量,以提高收敛速度。通过对多聚焦图像进行融合仿真实验,结果表明算法收敛速度快,可得到图像融合的Pareto优解。
There are many image fusion evaluation methods, but they can not reach the best at the same time and even be incompatible. Image fusion can be regarded as an optimization process of several evaluation factors. Image fusion schemes based on optimization with improved gene expression programming are proposed. Decision making with multi region features is used to reduce solution space for search. This improves the quality of initial swarm and increases convergence rate of optimization. Experiments of muhifoeus images show that it can get Pareto solutions quickly and have better convergence rate.
出处
《宇航学报》
EI
CAS
CSCD
北大核心
2009年第4期1658-1662,共5页
Journal of Astronautics
基金
国家自然科学基金资助(60672140)
教育部新世纪优秀人才支持计划资助(NCET-05-0912)
关键词
图像融合
多目标优化
基因表达式编程
Image fusion
Multi objective optimization
Gene expression programming
