高山冰川遥感提取方法研究

遥感的应用使得对冰川大尺度全覆盖、多时相变化的监测成为可能,然而冰川信息遥感提取方法的误差大等难题成为影响冰川监测的障碍。本文综合分析比较了目前已有的多种冰川提取方法的有效性,得出提取冰川范围精度最高的是面向对象的目视判读方法,其次是最大似然法监督分类、面向对象的自动分类、比值阈值法、雪盖指数法等。各自动方法提取冰川面积均有较大误差,且误差主要出现在冰舌末端、阴影区、薄冰区和云层遮盖范围等区域。本文将面向对象的目视判读法应用于冰川提取中,在保证信息提取精度的同时提高了传统解译的效率。

Study on Mountain Glacier Extraction Methods Using Remote Sensing Data

The application of remote sensing data makes it possible to monitor glaciers change in complete coverage of large scale and multi temporal.However,errors in glacier extraction methods using remote sensing data have become the chief obstacles to monitor glaciers change.This paper has a comprehensive analysis and comparison on the existing glacier interpretation methods.Through research,this thesis puts forward main conclusions as follows:Object-oriented visual interpretation method has the maximum accuracy in extracting glacier distribution in the experimental images,followed by supervised classification-maximum likelihood,object-oriented automatic classification,ratio threshold method,snow cover index method and so on.The computer automatic interpretation methods lead to relatively large errors.And the errors mainly appear in the end of glacier tongues,shadow areas,thin glaciers and cloud cover areas.The object-oriented visual interpretation method presented in this paper can solve the main problems in the automatic classification,guarantee accuracy and improve the efficiency of traditional visual interpretation at the same time.