多元信息支持的岩溶区土地利用信息提取方法构建

Land Use Information Extraction Method Based on Multi-information Support in Karst Area

土地利用信息是重要的基础数据,其准确程度将对其后续应用工作的结果产生重要影响。岩溶区地形复杂、地表覆盖类型多样,使用目视解译和传统监督分类的方法都存在一定的局限。通过结合遥感影像和植被指数、湿度指数、地形等多元信息,结合主成分分析方法,利用决策树实现上述多元信息建模,进而以实地调查样点为目标函数获取决策树的分类决策参数阈值,从而构建了多元信息支持的岩溶区土地利用信息提取方法。经试验区的模型应用和样点验证表明,该方法获取的土地利用信息基本符合典型岩溶区土地利用比例和分布特征,总体分类精度达到82%。该方法能够加深对多维遥感信息的认识,充分挖掘出其中隐含的信息,并根据这些信息提高岩溶区土地利用信息提取的精度,尤其能体现出岩溶区地貌特征对土地利用空间分布的影响。

Land use information is the important basic data,whose accuracy will have a significant impact on the subsequent researches.Given the complex terrain and heterogeneous land cover types,there are certain limitations in using visual interpretation and supervised classification methods to obtain land use information in karst areas.In this paper,we coupled remote sensing images with vegetation index,humidity index,terrain and other information,in combination with principal component analysis method.The process mentioned above is conducted through the decision tree modeling method.Decision parameter threshold of the decision tree is obtained by taking field samples to formulate the objective function.Then the land use information extraction method under the multi-information support in karst area is built.The results show that land use information obtained is in line with the characteristics of proportion and distribution of land use in typical karst areas,and overall classification accuracy reached up to 82%.This method can deepen the understanding of the multi-dimensional remote sensing information,fully tap the implied information,and improve the extraction accuracy of the land use information in karst area as well.In particular,it can reflect the influence of topographic features on the spatial distribution of land use in karst area.