摘要
针对机载LiDAR点云信息挖掘研究,提出了改进层次分析的机载LiDAR点云分类方法。首先选出5个判定准则,使用层次分析方法对每一个点云做出相应的判定,然后根据一定的规则对判定的结果生成一个二进制信号,最后使用BP神经网络方法确定判定因子在不同地物分类中的权重并对机载LiDAR点云分类。实验数据结果表明:该方法能够很好的将点云数据分类为高大的树、人工建筑物、低矮植物、地表和道路,分类精度Kappa系数达到0.87。
For the study on information mining of airborne LiDAR point cloud,this paper proposes airborne LiDAR point cloud classification based on improved analytic hierarchy process(AHP).Firstly,five criteria are selected to decide each point cloud by means of improved analytic hierarchy process,and then a binary signal is generated for the results according to certain rules.Finally,the weights of the decision factors in different ground objects are determined by BP neural network method and the airborne LiDAR point clouds are classified.The experimental results show that this method can well classify the point cloud data into tall trees,artificial buildings,low plants,land surface and roads,with classification accuracy Kappa coefficient up to 0.87.
作者
李晓天
姜刚
张玉
Li Xiaotian;Jiang Gang;Zhang Yu(School of Geology Engineering and Geomatics,Chang’an University,Xi’an 710064,China;School of Earth Sciences and Engineering,Hohai University,Nanjing 210098,China)
出处
《甘肃科学学报》
2019年第1期86-91,共6页
Journal of Gansu Sciences
基金
国家自然基金资助项目(41501498)
中央高校基本科研业务专项资金资助(300102268207)
关键词
机载LIDAR
改进的层次分析
人工神经网络
分类
Airborne LiDAR
Improved analytic hierarchy process
Artificial neural network
Classification
