Least Squares Support Vector Machines for Aerial Images Segmentation

SVM has desirable classification ability even if with fewer samples. In addition, LS-SVM reduces the complexity further through replacing the inequality of SVM by equality. This paper applies LS-SVM to aerial images segmentation. This paper researches on the different kernel and sparse of LS-SVM. The kernel influences aerial image segmentation. The briefness of decision function is reached by the LS-SVM sparseness. The experiments show the segmentation results of LS-SVM are better.