Study On Automated Extraction Methods Of Highway Elements From Mobile Lidar Point Clouds

With the rapid development of the smart city,the market put forward new requirements for data acquisition equipment from the data acquisition speed,precision,efficiency,interference factors,vehicle-borne laser scanning technology therefore become a research focus.Vehicle-borne laser scanning technology has the characteristics of non-contact,high precision,high speed,wide range and intuitive results.It has been used in the field of urban planning,inspection of road and railway and so on.Although the vehicle-borne laser scanning technology can provide enough commercial value in theory,because of its high cost,huge data,poor efficiency of data analysis and data application,resulting in low utility ratio of technology.The study,starting from the market demand,mainly research on how to automatic,efficient and accurate extraction the highway road elements of need from vehicle-borne laser scanning data,so as to promote the development and the popularity of vehicle-borne laser scanning technology.This paper introduces the recent development of laser radar technology and vehicle-borne laser scanning technology developed on the basis of laser radar technology,and summarizes the current situation of system integration and data post processing about vehicle-borne laser scanning technology.Basing on these research and relying on our equipment resources,a set of vehicle-borne laser scanning system is integrated,and its working principle and source of error is analyzed,and the main parameters of error are calibrated.Finally,precision of the results by vehicle-borne laser scanning system is analyzed.Be sure of precision of the results by vehicle-borne laser scanning system after calibration meet the application requirements,then point cloud of highway by vehicle-laser system is analyzed,and an automatic,efficient and accurate method to extraction elements from point cloud is put forward,to facilitate the planning and maintenance of highways,maintenance of roads and ancillary facilities.In this paper,the extraction methods of road elements are briefly described as follows:For the extraction of pavement point.First,each scan line is extracted by the time interval of adjacent points.Then,center of scanner at the time each point on the scanning line is collected is projected on to a plane obtained based on the principle of integration of vehicle borne laser scanning system.The projection point is called the track point in pavement.Based on the analysis of the slope of the coordinates of each point on the scanning line and the track point in pavement at the corresponding time,the road point closest to the laser scanning system on the each scanning line is extracted.The point is called the starting point of pavement on the scanning line.According to the collected sequence,the slope and slope difference of adjacent points on the scanning line are analyzed.If the slope and slope difference of point changes remarkably,the point may be the boundary point and is confirmed by analysis point slope and slope difference of the last point and the next point.Finally boundary end point of pavement on the scanning line is extracted.The continuity between the starting point and the boundary point of the scanning line is regarded as the final point cloud of pavement.To extract points of road marking,the point cloud of pavement has been extracted from the cloud is extracted by reflectance filtering.Then,to remove outliers through the method of dividing three dimensional grid by 4 times of point spacing.For the extraction of the guardrail,first of all,based on the characteristics of the boundary point of the scanning line is adjacent to points of the guardrail,guardrail point cloud with noise point is divided.Because the main body of each scanning line in guardrail point cloud with noise point is relatively concentrated and the clutter is relatively discrete,by using the method of extracting the punctate and linear dense areas on the different projection plane to extract the points of the main pillars of the guardrail and the point cloud of the wave barrier plate,and merge them to get the point cloud of guardrail.For the extraction of the street lamp,first of all,based on the characteristics of the boundary point of the scanning line is adjacent to points of the street lamp,the light pole of street lamp is divided.Through the method of extracting point above the pavement to extract the lamp cap of street lamp is divided.And merge them to get guardrail point cloud with noise point.Because main body in the initial segmented results is relatively independent,to divide super voxel using clustering method and features of super voxel are computed.To merge axially adjacent super voxel into a target.To count the voxel features contained in each target,and a target includes both horizontal plane and vertical rods is considered as a street lamp.According to the distribution characteristics of road elements to segment the preliminary results from original point cloud can reduce the amount of data for process;according to the geometrical feature of road elements to extract point cloud can ensure the accuracy of extracting.It is proved that the method is simple and efficient,which greatly improves the practicability of the laser scanning system.