摘要
以蓬莱市为研究区,对TM影像经过预处理后,进行典型地物的光谱分析。在此基础上,运用决策树分类法,选取影像的光谱特征值、NDWI值、NDVI值、K-T变换信息和DEM值等数据作为测试变量,选择适当阈值设定判别规则,建立决策树模型进行土地利用覆盖信息提取并做出精度评价。将其提取精度与监督分类结果精度进行比较,结果表明其分类精度有很大提高,尤其在研究区是丘陵地形的情况下,DEM数据的使用使林地、果园的可分性大大加强。
In this paper, taking Penglai as a study area, after the TM images of this area are preproeessed, spectrum analysis of typical surface features are carried out. On these basis, by using decision tree classification , selecting spectral characteristics, NDWI, NDVI, K - T Transformation and DEM data as test variables, using proper thresholds for setting discriminating rules, a simple decision tree model is built. On the basis of these discriminating rules for extracting land covering information, an accuracy assessment is given to the result by stratified random samples. It is proved that the decision tree classification can get higher accuracy result compared with the supervised classification method. Especially, in the case of hilly topography in this area, by using DEM datas to classify model for distinguishing woodland and orchard, the result is more accurate.
出处
《山东国土资源》
2009年第11期52-56,共5页
Shandong Land and Resources
