摘要
利用Nave Bayesian网络的学习和推理机制,提出一种在遥感影像上提取居民地目标的方法。该方法通过对所选取的正负样本进行学习,获取Bayesian网络的重要参数,即条件概率和概率分布密度。在此基础上,根据正负样本所构建的条件概率网,对未知类别信息的影像进行分类,从而获取居民地目标的信息。通过对实际全色SPOT5影像中居民地目标的提取,表明该方法具有较高的识别率。
The method is proposed to extract the images based on Naive Bayesian network. In this residential areas on panchromatic remote sensing procedure, the conditional probabilities and prior probabilities are obtained from learning the positive and negative samples. On the basic, unknown regions that have not the category information will be classified into residential areas and non-resi- dential areas by learning probability network. Experimental results of extracted residential areas on the images of SPOT5 are presented to illustrate the merit and feasibility.
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2007年第12期1103-1106,共4页
Geomatics and Information Science of Wuhan University
基金
国家科技支撑计划重点资助项目(2006BAB10B01)
