An Algorithm on Extraction of Saline-Alkalized Land by Image Segmentation Based on ETM~+ Image

摘要

Based on ETM+ remote sensing image, according to geographical distribution and spectral characteristics of saline-alkalized land, especially that of light, and moderate saline-alkalized land, an algorithm of semi-supervised multi-grade mean weighed seed region searching is presented. Experimental results indicate that the algorithm has high operation speed, high segmentation accuracy and certain fault tolerance, and also, to a great extent, over-segmentation problem can be perfectly solved. But because of mixed distribution of sand land and saline-alkalized land, they can not be effectively segmented by the algorithm.