The study of high resolution satellite image classification based on Support Vector Machine

This study chose Quick Bird satellite image with high resolution and spatial information as the resource origin of image classification and used Support Vector Machine (SVM) to achieve the goal on classification. We present two of spatial information which are Principal Component Analysis (PCA) image using 2-D discrete wavelet transform (DWT), and image segmentation. The DWT is used to generate spatial images from individual wavelet subbands. These feature vectors combined with original spectral image are first used for training and later for testing the SVM, decision tree, and maximum likelihood classifier. The proposed method produces promising classification results for spatial information analysis problems.