Using iterative weighting to fit the circular parameters of vertical cylinder in point clouds

Abstract

Most spatial data collection and extraction have transited from field work to office work by surveying and mapping technical innovation. More and more studies focus on the 3D object modeling based on automatic and semi-automatic modeling algorithms. Especially, the point clouds with highly density and precision collected from LIDAR (light detection and Ranging) are beneficial for 3D modeling by different algorithms. Therefore, the objective of this study is to develop the fitting algorithms to determine the circular parameters of the vertical cylinders of streetlamps or traffic poles in road environment from the point clouds. Unfortunately, two reasons make the fitting more difficult due to the real scanning situations. The first one is incomplete scanning because of the scanning position. Therefore, the point clouds cannot cover all the surface of the vertical cylinders. The second one is that not all the point clouds are from surface of the vertical cylinders, part of them are reflected from other attached objects, such as advertising flags or traffic sign devices. To cope with the second problem, least square with iterative weighting is used in this study to overcome the difficulties that we just mentioned about. Finally, the fitting results will be discussed for 3D modeling of the vertical cylinders in the future study.