TY - GEN
T1 - TriSpaSurf
T2 - 3rd Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2020
AU - Zheng, Hanfeng
AU - Di, Huijun
AU - Han, Yaohang
AU - Gong, Jianwei
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - Vehicle is one of the important subjects studied in the domain of computer vision, autonomous driving and intelligent transportation system. 3D models of vehicles are used widely in the literature of vehicle categorization, pose estimation, detection, and tracking. However, previous work uses only a small set of 3D vehicle models either from CAD design or multi-view reconstruction, limiting their representation ability and performance. A feasible approach to acquire extensive 3D vehicle models is desired. In this paper, we are interested in 3D surface reconstruction of on-road vehicles from sparse point cloud captured by laser scanners equipped ubiquitously on autonomous driving platforms. We propose an innovative reconstruction pipeline and method, called TriSpaSurf, which could reconstruct unbroken and smooth surface robustly from just a single frame of noisy sparse point cloud. In the TriSpaSurf, triple-view 2D outlines are first fitted on the 2D points from the projection of 3D point cloud under each view, and then 2.5D surface reconstruction is carried out under the guidance from triple-view outlines. By projecting 3D point cloud onto 2D views, 2D outlines could be estimated robustly due to the reduced complexity and higher signal-to-noise ratio in 2D views, and could provide fairly stable and tight multi-view constraints for 3D surface reconstruction. The effectiveness of our method is verified on the KITTI and Sydney dataset.
AB - Vehicle is one of the important subjects studied in the domain of computer vision, autonomous driving and intelligent transportation system. 3D models of vehicles are used widely in the literature of vehicle categorization, pose estimation, detection, and tracking. However, previous work uses only a small set of 3D vehicle models either from CAD design or multi-view reconstruction, limiting their representation ability and performance. A feasible approach to acquire extensive 3D vehicle models is desired. In this paper, we are interested in 3D surface reconstruction of on-road vehicles from sparse point cloud captured by laser scanners equipped ubiquitously on autonomous driving platforms. We propose an innovative reconstruction pipeline and method, called TriSpaSurf, which could reconstruct unbroken and smooth surface robustly from just a single frame of noisy sparse point cloud. In the TriSpaSurf, triple-view 2D outlines are first fitted on the 2D points from the projection of 3D point cloud under each view, and then 2.5D surface reconstruction is carried out under the guidance from triple-view outlines. By projecting 3D point cloud onto 2D views, 2D outlines could be estimated robustly due to the reduced complexity and higher signal-to-noise ratio in 2D views, and could provide fairly stable and tight multi-view constraints for 3D surface reconstruction. The effectiveness of our method is verified on the KITTI and Sydney dataset.
KW - 2.5D representation
KW - Surface reconstruction
KW - Triple-View outline
UR - http://www.scopus.com/inward/record.url?scp=85093821279&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-60633-6_33
DO - 10.1007/978-3-030-60633-6_33
M3 - Conference contribution
AN - SCOPUS:85093821279
SN - 9783030606329
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 398
EP - 409
BT - Pattern Recognition and Computer Vision - 3rd Chinese Conference, PRCV 2020, Proceedings
A2 - Peng, Yuxin
A2 - Zha, Hongbin
A2 - Liu, Qingshan
A2 - Lu, Huchuan
A2 - Sun, Zhenan
A2 - Liu, Chenglin
A2 - Chen, Xilin
A2 - Yang, Jian
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 16 October 2020 through 18 October 2020
ER -