摘要
以福建省顺昌县为研究区,首先利用光谱、纹理等信息对森林进行初分类;在此基础上,分析各森林类型在不同地形条件下的混淆情况,借助GIS手段,建立再分类规则,实现森林分类精度的提高。从分类总精度看,再分类结果较初分类结果高出9.11%,而Kappa系数则高出0.134 8,说明GIS支持下的决策树分类可较大幅度地提高南方山地丘陵区域的森林分类精度,具有良好的应用前景。
Taken Shunchang County in Fujian Province as the study area,the paper first made use of the spectral and textural information to realize the preliminary forests classification.On the basis of this,it analyzed the confu-sion condition of these forest types in different terrain conditions,and established further classification rules sup-ported by GIS,so as to increase the forest classification accuracy.The overall classification accuracy of further classification result was 9.11% higher than the preliminary one and the overall Kappa statistics was 0.134 8 high-er,which indicated that Decision Tree Classification supported by GIS could increase the forests classification ac-curacy a lot in mountain areas of Southern China a lot and had a good application prospect.
出处
《北京联合大学学报》
CAS
2011年第1期34-40,45,共8页
Journal of Beijing Union University
基金
国家自然科学基金(40971043)
福建省自然科学基金(2008J0117)
关键词
地理信息系统(GIS)
山区
遥感
决策树分类
森林类型
Geographic Information System
mountain areas
remote sensing
decision tree classification
forest types
