We proposes a novel and robust 3D object segmentation method, the Gaussian Density Model (GDM) algorithm. The algorithm works with point clouds scanned in the urban environment using the density metrics, based on existing quantity of features in the neighborhood. The LiDAR Velodyne 64E was used to scan urban environment.