Comparison of Different Remote Sensing Monitoring Methods for Land-Use Classification in Yunnan Plateau Lake Area

Abstract

Remote sensing image classification is an important technology to get information. At present, different remote sensing monitoring methods has been widely used in region land cover. To improve classification accuracy is the key of remote sensing data processing and application. This paper selects Xingyun Lake that the typical Plateau Lake area of Yunnan province and the surrounding lakeside zone as research area. Based on the 30 TM Landsat remote sensing image of the research area, using supervised classification, BP neural network, and object-oriented classification to compare accuracy of three kinds of classification methods. It was found that development of BP neural network and object-oriented classification training produces more accurate results than supervised training. Object-oriented classification also produced more accurate classification than the BP neural network classification, but did not improve the accuracy significantly. The results will help to promote surface coverage information of remote sensing rapidly extraction and dynamic monitoring in the Yunnan plateau lake, moreover, it has important scientific significance to protect and formulate rationalization.