Urban road marking extraction method based on vehicle laser point cloud
Aiming at the reflection intensity characteristics of road markings from vehicle-mounted laser point cloud,an urban road marking extraction method based on vehicle laser point cloud is proposed.Firstly,a ground filtering method combining cloth simulation filtering and elevation skewness balance is presented to take advantage of the adaptability of skewness balance filtering to eliminate the problem of low vegetation remaining after cloth filtering.Then,the normal vector density clustering is adopted to extract the road point cloud,and the road point cloud is converted into an intensity feature map by inverse distance weighted interpolation.In order to alleviate the jaggedness extracted from the markers,fast bootstrap filtering is introduced to smooth the edge information of the road markers.Finally,the maximum entropy threshold segmentation and morphological ratio filtering are used to refine the road marker.The experimental results show that the method can effectively extract the road marking point cloud,the average recall rate of extraction is 80.98％,the average accuracy rate is 96.89％,and the average comprehensive evaluation index is 88.19％and it can use the road marking point cloud intensity information to extract the road marking point cloud in a more complete way.
point cloudcloth simulation algorithmskewness balancing algorithminverse distance weightedfast guided filtering
NETL
NSTL
万方数据
