Research on effect of prior probability for maximum likelihood classification accuracy in remote sensing image analysis

摘要：

　　Maximum likelihood classification is classic and widely used in remote sensing image analysis The mathematical theory foundation is based on Bayesian formula for the minimizing Bayesian error.Many researches have indicated that using prior probability can make Maximum likelihood classifier perform better,so the correct setting of Prior probability does improve the accuracy of Maximum likelihood classifier.In this paper,Firstly,Based on the Bayesian formula is deduced from the Maximum likelihood discriminant function.Secondly,the method of determining prior probability for Iterative loop algorithm is shown.Finally,an experiment about Songhuaba reservoir in northwest of Kunming,Yurman province in China,which is carried out by Iterative loop algorithm verify and explain the effect of Prior probability on Maximum likelihood classification,In comparison with the first classification,then after the sixth classification,Overall Accuracy and Kappa coefficient were improved by 2.04% and 0.0543 respectively.

关键词： maximum likelihood classification prior probability iterative loop algorithm Bayesian formula

作者: X.Zhu H.J.Duan J.M.Chen J.Li

作者单位: Yunnan Environment Science Insititute, Kunming, China Department of Physics, Faculty of Basic Science and Technology, Kunming, China Kunming University of Science and Technology, Kunming, China

会议类型: 国际会议

会议名称: 2012 International Conference on Frontiers of Energy and Environmental Engineering(2012年第二届能源与环境工程前沿国际会议(ICFEEE2012))

会议地点: 香港

会议语种:英文

页码: 166-171

在线出版日期: 2012-12-01（万方平台首次上网日期，不代表论文的发表时间）