【摘要】 在分析当前滑坡灾害信息识别方法的基础上,指出这些方法的适用性和不足之处,并针对这些不足之处,以滑坡的地学原理为依据,提出了以多时相遥感影像为数据源,结合纹理分析的变化检测自动识别滑坡灾害信息的方法。最后,以地震前(2006-5-14)后(2008-6-3)的北川县城及附近区域的福卫2号多光谱遥感影像作为数据源,以结合比率变换和纹理分析的变化检测进行滑坡灾害信息识别试验。结果表明,结合纹理分析的变化检测方法能够突破影像光谱特征的局限,对于滑坡灾害信息识别具有显著的适用性。更多还原

【Abstract】 The current recognition approaches of landslide disasters mostly are based on visual interpretation,which may result in low productivity.This paper analyzes the current identification methods of the landslide disasters,and their applicability.According to the disadvantages and principles of landslide disasters,we bring forward a more effective approach to identify landslides,which uses change detection combined with texture analysis in multi-temporal images.Additionally,this paper takes Beichuan county which has been affected by the great earthquake seriously as a case to implement the new method in application of landslide automatic identification,and takes multi-temporal ROCSAT-2 as the data resources of pre-landslide and post-landslide study respectively.The finding shows that the method combined with change detection can use the more image information including texture,and it is available to the automatic recognition of landslide disasters.更多还原