Abstract：According to the characteristics of the road in vehicle-borne LiDAR point cloud data, analgorithmofk-means clustering was proposed based on the trajectory, aiming to research the automatic extraction method on road boundary. Thealgorithm is described as followings. firstly, simplify the complex trajectory into one single trackfordata preprocessing; then, insert sections using the trajectory, getting section lines through projection in cross section of the point cloud; thirdly, use K-means clustering algorithm to extract the road boundary based on the section line; lastly, check and optimize the result for accurate road boundaryinformation. The test resultshows that the algorithmcan automatically extract road boundary efficiently and accurately.

Keyword：vehicle-borne LiDAR;trajectory auxiliary;point cloud;refining extraction;K-meansclustering;