Extracting target information in multispectral images using a modified KPCA approach

Abstract

In this paper, a modified kernel principal component analysis (KPCA) approach is proposed to extract target information in multispectral images. The advantage of this method is to be able to extract target information with fairly small computation complexity compared to the standard KPCA when a large number of input samples need to be processed. Finally, some experimental results demonstrate that our proposed approach is effective and efficient for analyzing and interpreting the multispectral images. © Springer-Verlag 2004.