The Weak Information Extraction and Application from Remote Sensing Satellite Images

摘要

Principal Component Analysis (PCA) is a powerful tool to extract structure information from high-dimensional data set. After PCA, most information of the data set is distributed on several preceding components. But in some applicants, such as remote sensing mine detection, the non-principal components, called weak information, are focused because they present mine erosion information. On some certain occasions, classical PCA is not enough to do the weak information extraction work. In this paper, Kernel PCA is proposed. By mapping the input space to the high-dimensional feature space, Kernel PCA can be considered as nonlinear PCA in the input space. So, it can extract weaker components. This paper mainly concentrated on the applicant of Kernel PCA in remote sensing mine detection.