面向对象的遥感湿地植被分类与信息提取——以微山湖为例

Object-oriented Remote Sensing Wetland Vegetation Classification and Information Extraction

在对面向对象遥感分类方法的基本原理与基本算法阐述的基础上,以山东省微山湖湿地植被分类为实验区数据,进行湿地植被分类训练和精度评定。结果表明,实验区居民区、水域、陆生植被、水生植被的面向对象的分类精度分别为:92.81%、93.78%、96.29%、95.38%,明显高于监督分类的分类精度92.13%、88.36%、89.24%、86.55%,充分展示出在湿地植被分类中面向对象遥感分类的优势。

Based on elaborating the basic principle and algorithm of object-oriented remote sensing classification method,with Weishan Lake wetland vegetation classification in Shandong Province as experimental data,the classification training and precision evaluation of wetland vegetable were carried out.The results showed that the object-oriented classification precision of residential area,water area,terrestrial vegetation,aquatic vegetation are 92.81%,93.78%,96.29%,95.38%,obviously higher than that of supervision classification 92.13%,88.36%,89.24%,86.55%,which fully display the advantages of object-oriented remote sensing classification in wetland vegetation.