利用动态贝叶斯网络进行多时相遥感变化检测

Multi-temporal Remote Sensing Change Detection Using Dynamic Bayesian Networks

利用动态贝叶斯网络(DBNs)在处理不同时相遥感数据时可以一次性输入多个时间段的数据,同时完成分类和建立输出类别之间的关联。采用北京东部地区1994年、2001年和2003年5月份Landsat TM遥感数据进行实验,实验结果表明:基于DBNs的变化检测方法是遥感变化检测的一种新的有效方法,在遥感时序数据动态变化分析的研究方面也展示了巨大的发展潜力。

Utilizing Dynamic Bayesian Networks （DBNs） to deal with multi-temporal remote sensing data, the multi-temporal data of different time can be input simultaneously, and the classification and the acquirement of relationships between the output types can be finished simultaneously. Using the Landsat TM remote sensing data of Beijing eastern area acquired in May of 1994, 2001 and 2003 for the experiment, the experimental results indicate that the DBN-based change detection method is a new effective method of remote seusing change detection, and show its great potential for the research on the analysis of the dynamic changes of remote sensing time-series data.