中文摘要：

RBF-SVM是场景分类中主流的分类算法之一,C和γ是其优化参数。本文分析了场景分类中RBF-SVM优化参数的形态分布,从实验结果发现场景分类中RBF-SVM参数优化属于多峰值的优化问题。参数与分类精确度之间不具有显式的函数关系,并且最优解的搜索空间很大,用传统的网格算法和枚举法很难满足需求。演化算法具有自组织、自适应、自学习等智能特征,是解决场景分类中RBF-SVM参数优化的有效途径。

英文摘要：

RBF-SVM is one of the mainstream classification algorithms in scene classification with C and γ as its optimization parameters.This paper analyzes RBF-SVM optimization parameters＇ morphological distribution in scene classification.The experiment results show that RBF-SVM parameter optimization in scene classification is a multi-peak optimization problem.Since RBF-SVM parameters do not have the explicit function relationship and the solution space is very large as well,so intelligent evolutionary algorithms should be used for RBF-SVM parameters optimization.