摘要
运用面向对象分类法中的基于监督分类和基于规则的滑坡识别方法,选择合适的特征属性,利用Aster和Geoeye的融合影像对构林坪流域进行滑坡信息提取,并对分类结果进行精度评价和比较.结果表明:基于监督分类的滑坡信息提取总体精度为66.58%,Kappa系数为0.65,具有较高的分类精度;基于规则的滑坡信息提取方法也取得了84.7%的识别结果,但是区域特殊地形地貌和引发滑坡因子的复杂性导致了72.6%的分歧因子.总体上基于面向对象分类法的高分辨率遥感滑坡信息提取在白龙江流域具有良好的适用性.
Through a selection of appropriate attributes, two object-oriented classification methods (supervised classification method and rule-based method) were applied to extract landslide morphological information for Goulinping Basin based on the fused images of Aster and Geoeye. The accuracy of the classifications were evaluated and compared and the results showed that the accuracy of landslide information extraction based on the supervised classification is 66.58%and the Kappa coe?cient is 0.65. The recognition accuracy based on the rule-based method is 84.7%, but the difference of factor is 72.6% due to the special topography and the complexity of the influencing factors of the landslide. In summary, the two classification methods have good applicability in Bailong River Basin.
出处
《兰州大学学报(自然科学版)》
CAS
CSCD
北大核心
2014年第5期745-750,共6页
Journal of Lanzhou University(Natural Sciences)
基金
国家科技支撑计划项目(2011BAK12B06)
国家自然科学基金项目(41172328)
甘肃省科技重大专项计划项目(1102FKDA007)
国家重点基础研究发展计划(973计划)项目(2014CB744703)
关键词
面向对象
滑坡识别
监督分类
遥感
object-oriented
landslide information extraction
supervised classification
remote sensing