基于改进的FCM和KPCA的多光谱图像特征提取方法

Feature Extraction for Multispectral Images Based on Improved FCM and KPCA

分析了PCA和KPCA对于提取多光谱图像非线性特征的不足,提出了一种基于改进的FCM和KPCA的多光谱图像特征提取方法。首先利用改进的FCM进行聚类分析,然后将获得的聚类中心作为输入样本,进行KPCA,从而得到主成分图像。试验结果表明,文中提出的方法具有良好的特征提取性能,可有效地提取多光谱图像的非线性特征。

The drawbacks of the standard principal component analysis （PCA） and the kernel principal component analysis（KPCA） in extracting nonlinear feature of multispectral images are analysed. A new method of feature extraction using improved FCM and KPCA is proposed. After clustering analysis by the improved FCM, the obtained cluster centers as input samples is used and then the principal component images can be obtained based on KPCA. The experiments show that the proposed method has effective performance in nonlinear feature extraction of multispectral images.