At present, the point clouds-based 3D target detection technology has many researches and applications in the field of car automatic driving, but there is no 3D target detection technology for ships. Ship navigation is different from car driving. There will be waves in the channel, undulations occur when the ship is sailing, and there is often fog above the channel, which affects the quality of lidar imaging. These will bring many difficulties to the detection of 3D targets of the ship. In order to deal with the complex noise in the point clouds in the waterway scene, we fused a variety of filtering algorithms to preprocess the point clouds, and in view of the ups and downs of the ship during the sailing process, a point clouds fusion sampling algorithm is introduced on the basis of the PointRCNN algorithm, and established lidar standard coordinate system. The experimental results show that the improved algorithm is more suitable for ship detection than the PointRCNN algorithm, and has higher detection accuracy for ship targets. This research expands the scope of application of 3D target detection, and provides a theoretical basis for subsequent applications such as ship gate risk warning.
Research on ship 3D target detection method based on LiDAR point clouds
International Conference on Mechanisms and Robotics (ICMAR 2022) ; 2022 ; Zhuhai,China
Proc. SPIE ; 12331 ; 123310E
10.11.2022
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Research on ship 3D target detection method based on LiDAR point clouds
British Library Conference Proceedings | 2022
|Detection of Cars in Mobile Lidar Point Clouds
IEEE | 2018
|