Research progress and prospect of LiDAR point cloud feature line detection and structuring
Three-dimensional(3D) real scene construction is an important part of the national new fundamental surveying and mapping construction,which is the construction of geographic entities that carry structure,semantics,human-machine compatible understanding,and real-time perception of Internet of Things(IoT) in 3D geographic scenes. Structuring is the process of segmenting individual geographic entities and obtaining their geometric contours and composition structures. With the reduced collection cost of large-scale and city-level laser point cloud data,the application of 3D point cloud is also becoming increasingly widespread.Based on the 3D point cloud,it is possible to achieve a structured representation of ground objects,involving the reconstruction of line,surface and body,semantic segmentation and instance segmentation,etc. The core content of point cloud data processing is the detection of point cloud feature line,feature line segmentation,and feature line structuring. Therefore,point cloud feature lines have theoretical and practical value for point cloud processing and industry applications. It is of great significance to conduct adaptive feature line detection and fine structured research on point cloud in urban scenes. The study of intelligent perception theory and structured expression methods for point cloud feature lines can provide a new perspective for point cloud data processing,which has important theoretical guidance for various engineering practices. It can serve specific applications such as 3D real scene construction,two-dimensional(2D) and 3D high-definition map generation. This paper summarizes the current research status at home and abroad from three perspectives,including the detection of the point cloud feature line,the segmentation of the point cloud feature line,and the structuring of the point cloud feature line.It concludes the problems in the detection of point cloud feature line and structuring research and determines the future research direction.
laser point cloudfeature line detectionfeature line segmentationfeature line structuringthree-dimensional (3D) real scenenew fundamental surveying and mapping
万方数据
维普
