Airborne light detection and ranging(LiDAR) point cloud data can provide the framework and fundamental technical support for many industry applications. Point cloud data is an important source of geospatial data for the construction of smart cities and the construction of three-dimensional(3D) real scene China. High-quality point cloud classification can significantly improve the solid 3D representation of geospatial data. Therefore,it is particularly important to summarize and sort out the research progress of airborne LiDAR point cloud classification. This paper explored point cloud classification methods based on multi-source maps,features,neural networks and deep learning,and multi-modal data utilization. This paper summarized the technical advantages and potential problems of various methods,and it analyzed their development trends. In the LiDAR point cloud classification of complex urban scenes,multi-modal data coupling of global inference was conducted through embedding optical images,fusing multi-source map annotation information,and combining neural networks and deep learning methods. All this is to achieve the classification of airborne LiDAR point cloud of high efficiency,high precision,and high accuracy,which will be the direction that requires in-depth research in the future.
关键词
机载激光雷达/点云分类/神经网络/深度学习/多模态数据/点云语义化
Key words
airborne light detection and ranging(LiDAR)/point cloud classification/neural networks/deep learning/multimodal data/semanticization of point clouds
维普
万方数据