面向对象方法的复杂地形区地表覆盖信息提取

Land cover information extraction for complicated terrain regions via an object-oriented classification method

探讨在复杂地形区基于面向对象分类方法与中等分辨率的Landsat 8 OLI影像相结合能否获得较好的信息提取精度.结果表明:单波段标准差、多波段相关系数和最佳指数因子是选择波段的较优方案,可有效减少各波段间的信息冗余量.通过对研究区进行地理分区的方式建立每一地理子区的分类层次、设定分割参数及规则,根据地物类别的分布特征引入归一化植被指数、归一化建筑物指数、改进归一化差异水体指数、数字高程模型数据、坡度等与地类相关的专题数据,以提高复杂地形区分类精度;整个流域脑山区总分类精度最高,达88.33%,Kappa系数0.86.

This study was to explore whether a better accuracy for information extraction can be obtained by using the combination of object-oriented classification method and medium-resolution Landsat OLI images in the complex terrain areas.The results showed that the single band standard deviation,multiband correlation coefficient and optimum index factor were the optimal solution of band,which could effectively reduce information redundancy between bands.In order to improve the classification accuracy,the study area was divided into different subareas,and classification hierarchy,segmentation parameters and rules were established respectively for every subarea;at the same time,according to the spatial distribution of feature categories,some related thematic data such as normalization difference vegetation index,normalized difference built-up index,modified normalized difference water index,digital elevation model,slope etc.were introduced to help the classification.The total classifying accuracy of the whole study area was 88.33%and Kappa coefficient was 0.86.