| The shipborne LiDAR system can quickly and accurately detect the 3-dimensional information of the surface of watercourse,vegetation,facades and bridges on both sides of the river.The shipborne LiDAR data is characterized by high redundancy,massiveness,diversity of targets,uneven distribution of point density,serious lack and unstructured features.So it is of great research significance to process and apply shipborne LiDAR point cloud data.This paper focuses on the classification and extraction of shipborne Li DAR point cloud data,inside which the extraction of bank line and the classification and optimization of typical terrain are primarily studied.The main work and innovations are as follows:1.Based on the introduction of the development of shipborne Li DAR technology and the rapid acquisition of 3D point cloud data in inland rivers,this paper analyzes and summarizes the current research status and key technical issues of the bank line extraction methods and point cloud classification methods,which provides research ideas for shipborne LiDAR point cloud data classification and extraction.2.The sensor composition and working principle of the shipborne LiDAR system are introduced in detail.Compared with the point cloud data obtained by the mobile/airborne LiDAR system,the characteristics of the shipborne LiDAR point cloud data are analyzed in detail,and three data organization methods and two denoising methods for shipborne LiDAR point cloud data are given.3.A method for extracting precise river waterside line using shipborne LiDAR data is proposed.Firstly,point cloud data is coarsely grid-divided,by exploiting density characteristics,grid elevation shoreline characteristics,marginal characteristic and continuity characteristics of river waterside line,rough river edge can be quickly determined combining with connected region labeling algorithm and edge detection algorithm;Then,the coarse extraction results were fine grid partitioned,more accurate bank line point is extracted by using the local lowest elevation point of single grid network algorithm with lowest distance constraints.Experimental results show that fine and reliable river bank can be extracted by this method.4.Aiming at solving the problem of low efficiency of point cloud classification caused by massive nodes and undirected edges when using traditional high-order conditional random field model,a point cloud classification method based on multi-scale voxel and high order random fields is proposed in this paper.Firstly,multiscale voxel is used as a node of undirected graph to replace the mass of discrete point clouds and reduce the number of nodes and undirected edges;Then,the result of the supervoxel segmentation is applied as a higher-order cluster and an unsupervised distributed spatial context is designed as a higher-order cluster eigenvector to improve the classification result;Finally,combined with the constructed graph model and each order eigenvector,the classical high-order conditional random field model is implemented for automatic point cloud data classification.The experimental results show that the method proposed in this paper can effectively classify shipborne point cloud data and the classification efficiency of the high-order conditional random field point cloud classification model is improved by 7 to 16 times under the premise of ensuring the classification accuracy.5.A point cloud classification optimization method based on the geometric features of facade is proposed.The rules of classification for facade point are established by using the geometric features of the facade,and a multi-sub-area region growing algorithm and unsupervised classification method are separately designed to optimize the building facade points and wrong-division facade points.The experimental results show that two types of common misclassification problems in the classification result are effectively solved,and the classification accuracy of the point cloud data is improved. |
PDF Full Text Request