基于间隔Fisher分析的几种改进算法  

Several Improved Algorithms Based on Marginal Fisher Analysis

作　　者：朱振凤[1] 范丽亚[1] 

机构地区：[1]聊城大学数学科学学院,山东聊城252059

出　　处：《聊城大学学报（自然科学版）》2012年第4期5-10,共6页Journal of Liaocheng University:Natural Science Edition

基　　金：国家自然科学基金资助项目(10871226);山东省自然科学基金资助项目(ZR2009AL1006)

摘　　要：在间隔Fisher分析(MFA)的基础上,提出了三种改进算法:正则化MFA,目标函数为减式的MFA和基于零空间的MFA.为了检验算法的有效性和先进性,针对AVIRIS遥感图像进行了实验.实验结果表明,所提算法均优于MFA.Three kind of improved algorithms, based on marginal Fisher analysis （MFA）, are proposed in this paper, which are regularized marginal Fisher analysis, marginal Fisher analysis with subtractive objective function and marginal Fisher analysis based on the null space. Experiments on AVIRIS remote sensing image are made in order to test the validity and advantage of the methods presented in this paper. The experimental results demonstrate that our proposed algorithms are better than MFA.

关 键 词：正则化MFA 目标函数为减式的MFA 基于零空间的MFA AVIRIS遥感图像 

分 类 号：O186.11[理学—数学]