基于多级决策树分类的土地利用与覆盖信息提取

Extraction of Land Use and Land Cover Information Based on Multilevel Decision Tree Classification

结合Landsat-8遥感数据,采用多级决策树分类方案,利用归一化植被指数、波段比值、主成分分量等光谱特征参数并融合其他非遥感知识,对黄河三角洲地区土地利用与覆盖的信息展开了全面的提取、研究与分析,获得了该地区5个一级类、12个二级类地物的分布情况,分类总体精度93.88%,优于传统监督分类。同时采用聚类、分类叠加和人机交互等分类后处理操作以获得更贴近地面实际的制图效果,开展基于海岸线的缓冲区分析以获得各地物特别是距离海岸线10 km、20 km范围内地物类型的空间分布并完成相关制图与分析,为黄河三角洲地区滨海土地的利用与开发提供了数据支持。

The paper has performed a comprehensive study and analysis over the land use and land cover information of the Yellow River Delta using the method of multilevel decision tree classification and has gathered the information on land distribution of 5 level - 1 land cover types and 12 level - 2 land cover types by utilizing various spectral feature parameters such as normalized difference vege- tation index （NDVI）, band ratio and principle components, etc, as well as other ancillary knowledge. The overall classification accu- racy is 93.88% and superior than supervised classification. Post classification processing like clumping classes, overlay and man - machine interaction correction were implemented to obtain mapping results which are more closely aligned with actual situations. Buffer zone analysis was conducted to obtain the distribution information on different land cover types especially those within the 10 km/20 km buffer zone of coastline. Thematic maps were made in the last part, offering the decision - making processes for this area＇ s land use and land development with data support.