Remote Sensing Image Extraction and Precision Analysis for Alpine Wetland Based on Coupling Analysis of Multispectral Factor PCA and Decision Tree

The study was conducted to learn remote sensing information extraction in alpine wetland in Maqu, Gansu. The first principle component obtained by utilizing TM multispectral data, digital elevation data, GIMMS NDVIand principal component analysis (PCA) was regarded as the classification character. The decision tree module for wetland information extraction was built through analyzing data with spatial character, spectrum character and statistic. And then the decision tree module was compared with methods of unsupervised classification and maximum likelihood. The research indicates the classification method based on multi-featured decision tree is able to be used in thematic information extraction, and has a preferable applicability and bright prospect in research on alpine wetland.