Research On Extraction Of Road And Traffic Lines Based On Vehicle-borne Laser Scanning Data

With the rapid development of the geographic information industry and its wide application,the way of obtaining geographic information data becomes diverse.The vehicle-borne laser scanning system integrates three-dimensional laser scanner,GPS,digital camera,inertial navigation system(IMU)and other sensors.It can obtain high-precision and high-density point cloud of the road and the objects beside both sides among the vehicle driving process,which opens up a new way for the rapid acquisition,updating and automatic extraction of road data,and provides a reliable data source for the extraction of high precision traffic lines.Aiming to extracting the urban road and traffic lines using the vehicle-borne laser scanning system in this paper,the main works of this paper are as follows:1.The background and significance of research as well as the status of domestic and aboard research in using vehicle-borne laser scanning data for extraction of road and traffic lines is introduced.Then the working principle of the vehicle laser scanning system,the coordinate transformation of data and the characteristics of point cloud data are introduced.2.An method of urban road point cloud data extraction based on pseudo-scan line generation is designed.The vehicle laser scanning data were processed by block using the trajectory data.Then part of the point cloud within each block were projected onto the profile.As city road has curbs then the curb points were extracted from the pseudo scan line which was got from the profile point cloud.At last,cubic spline interpolation method was used to get the road boundary and the road point cloud is extracted using the road boundary.Experiments show that this method is feasible and effective.3.A generation method of laser scanning point cloud reference image based on eigenvalues calculation is designed.Based on the analysis of the theory and related applications of point cloud depth image,intensity image and CCD image,the point cloud data is firstly projected to the plane.Then,the grid feature value is calculated combined with the plane distance feature and the reflection intensity characteristic of the laser scanning point with each grid.Finally,the grid feature value is calculated to the gray space,which transforms the point cloud data in the three-dimensional space into a two-dimensional point cloud reference image.4.An traffic lines extraction method based on image binarization is introduced.Considering the characteristics of the target cloud image with clear target and less noise,the maximum interclass variance algorithm is used firstly to binarize the point cloud reference image.But there are small holes and gaps among the traffic lings in the binarization image.In view of the fact that the mathematical morphology closed operation has the characteristics of filling the smaller holes,the binarization image is then processed by the mathematical morphology closed operation,which resulting in the traffic lines is extracted completly.Finally,combing the pixel coordinated of the extracted traffic lines with the information of point cloud,the geographic coordinates of the key points of the traffic lines are calculated,which can be the auxiliary data for the high-resolution image aerial triangulation of the low-altitude remote sensing.