Automated extraction of urban roadside trees from mobile laser scanning point clouds based on a voxel growing method

This paper presents a new method for extracting urban roadside trees automatically from mobile laser scanning point clouds. This method mainly includes three steps. First, ground point clouds are removed by voxel-based upward growing method. Second, Euclidean distance segment method is used to cluster non-ground point clouds into certain individual objects. Then crown seeds of the initial layer is found by comparing the number of points in each layer after using the voxel modeling algorithm. Crown seeds in other layers can thereafter be detected via the upward inter-sectional analysis. Third, a crown voxel growing algorithm is used to make the crown grow in horizontal. The experimental results show that the voxel models of the individual roadside trees can be automatically and effectively extracted with our method.