Font Size: a A A

Research On The Extraction Of Mountainous Roads From LiDAR Based On Multi-feature Constraints

Posted on:2019-09-21Degree:MasterType:Thesis
Country:ChinaCandidate:D F HanFull Text:PDF
GTID:2370330548961235Subject:Engineering
Abstract/Summary:PDF Full Text Request
Road is a constituent element of spatial geographic information and also an integral part of human traffic.Accurate road information has great significance to all aspects of map updating,vehicle navigation and transportation management.With the development of economy and the progress of society,people's demand for road information is also constantly expanding.How to accurately and efficiently extract road information has become a research hotspot at this stage.As an active space measurement system,airborne Li DAR integrates laser ranging,GPS positioning,inertial navigation and other advanced technology.It can not only obtain the three-dimensional coordinates of the ground,but also record the number of echoes,reflections intensity and other information.The emergence and development of this technology provide a new way for road extraction.At present,the research on airborne Li DAR t echnology is mainly focused on the urban areas with flat topography.There are relatively few studies on extraction of roads in mountainous areas,and the traditional methods based on remote sensing images have been unable to meet people's needs in time.Taking the above factors into account,this paper systematically explores the extraction technology of mountainous road based on Li DAR point cloud data,and focuses on the filtering algorithm of Li DAR point cloud and the classification method of road points in mountainous areas.The specific research includes:1)Summarize the development history of airborne Li DAR technology and the research status at home and abroad.Expound the composition and working principle of airborne Li DAR system.Briefly analyze the characteristics of Li DAR point cloud data and introduce the application fields of airborne Li DAR system.2)Introduce the basic principle of point cloud filtering.Summarize the principles and implementation processes of several typical filtering algorithms of point cloud in detail,and analyze their advantages and disadvantages.Propose a filtering algorithm based on regular mesh generation strategy.The algorithm initially divides the huge amount of point cloud data into strips and selects the appropriate spacing to subdivide each strip into equidistant grid data.The grid data is used as the input to an iterative polynomial fitting process,after which the point cloud is classified based on a controlled threshold.Select the sample data provided by ISPRS to test the performance of the algorithm,the experimental results show that the proposed algorithm can quickly and efficiently classify data of mountains with different characteristics while retaining terrain feature information better than other algorithms.The average accuracy of recognition is greater than 92%.3)Review several current methods of road extraction based on Li DAR point cloud data.Analyze the typical characteristics of mountainous roads and propose an extraction method of mountainous roads based on multi-feature constraints.Firstly,use the neighborhood search algorithm of k-d tree to calculate and extract neighborhood points,normal vectors and roughness constraints of each discrete point.Set appropriate thresholds to obtain candidate road point sets.Next,adopt DBSCAN algorithm to extract road clusters and further eliminate the non-road point sets through cluster point number and shape index.Then use the morphological image algorithm to extract road centerline.Select the Li DAR data of a mountainous area in Hong Kong for experimental verification and compare with the satellite images,it is proved that the proposed method can extract the information of mountainous roads more completely,with high efficiency.Finally,point out the future research direction of mountainous roads extraction.
Keywords/Search Tags:LiDAR, mountainous roads, filter algorithm, mesh generation, feature constraints, k-d tree, DBSCAN
PDF Full Text Request
Related items