Multispectral Point Cloud Classification: A Survey

Point cloud classification holds paramount significance in contemporary remote sensing applications. And compared with traditional Light Detection and Ranging (LiDAR), Multispectral Light Detection and Ranging (MS-LiDAR) not only has the capacity to obtain richer point clouds but also detailed spatial location information and extensive spectral information. How to use multispectral point clouds to accomplish classification tasks has become one of the research hotspots for a variety of applications, such as land cover and forestry management, etc. It has attracted the attention of numerous researchers. In order to help more scholars understand the research in the field of multispectral point cloud classification, this paper provides a comprehensive analysis and summary of existing methods. The main work includes an in-depth study of image-based and point-based methods, while graph-based methods are discussed separately. In addition, this paper discusses the future prospects of multispectral point cloud classification techniques.