摘要： 选择了北京奥运主场馆及其周围的地区作为实验区,购置陆地卫星ETM+6个波段数据,从学习机制和技术流程上对贝叶斯网络分类和最大似然分类进行了对比,实验结果表明:贝叶斯网络分类方法在提高遥感数据的分类精度方面具有较大的研发潜力,贝叶斯网络为遥感数据分类处理提供了一种可选择途径。

关键词: 遥感数据, 贝叶斯网络分类, 最大似然分类

Abstract: Taking the Landsat TM data in Beijing Olympic games and the surrounding areas acquired in May 29th, 2003 as example image andwith the same study dataset, this paper classifies the image into four kinds of land use class separately using Bayesian network model and maximumlikelihood model. From the result of classification with two methods, accuracy of bare place and water are same by the large, and accuracy of urbanland and plat land can be seen, using the Bayesian network is higher than using maximum likelihood.

中图分类号: 

• TP75