摘要
为了精确提取地表覆盖类型复杂、水体类型多样的山区水体信息,本文提出一种基于Landsat 8多光谱影像的综合集成的决策树分类法。该方法以决策分类树为模型框架,以分层分类的遥感地学分析思想为主线,将自动水体提取指数(AWEI)以及基于知识的缓冲区图像分割法集成到决策树分类器中进行分类。对两个典型测试区域进行精度评价,得到Kappa系数分别为0.9596和0.8960,总体精度分别为99.7792%和99.6577%。
In this paper, we introduce one comprehensive integrated hierarchical classification methodology in which Automate Water Extraction Index （AWEI）, Normal Difference Three Band Index （NDTBI） and buffer zone segmentation image based on knowledge were integrated into decision tree classifier , with the object to pre- cisely extract various types of water from Landsat 8 multispectral image in the broad mountain areas with complex land cover classes. Accuracy assessment of two chosen two test sites shows that Kappa coefficients are separately 0_9596 and 0.8960 and the overall accuracies are 99.7792% and 99.6577%.
作者
饶萍
王建力
RAO Ping WANG Jianli(School of Geography Sciences, Southwest University, Chongqing 400715, China Guizhou University of Engineering Seienee, Bijie 551700, China)
出处
《贵州大学学报（自然科学版）》
2016年第5期17-24,共8页
Journal of Guizhou University:Natural Sciences
基金
国家自然科学基金项目(40971122)
中央高校基本科研业务费专项资金项目(XDJK2011C089)
关键词
水体
决策树
分层分类
中分辨率
遥感
LANDSAT
8
water
decision-making tree
hierarchical classification
medium resolution ratio
remote sensing
Landsat 8
