@inproceedings{2c7b950ee3a64a9ba40433a2aa421e58,
title = "A Robust Point Cloud Registration Method for Structured Scenes Based on Plane Elements Using Global Information",
abstract = "Point cloud registration is a necessary step in the fields of mapping and positioning when a complete surface description of the scene needs to be constructed. This paper proposes a robust point cloud registration method based on plane elements, which uses plane information in the environment to solve the transform matrix. A combination of region grow and Random Sample Consensus (RANSAC) algorithm is employed to segment the planes from the scene. A novel metric based both on the plane area and the number of plane points is utilized to filter abundant planes in the environment. Coincidence degree for diffenent candidate transforms are calculated after down-sampling to determine the optimal result of the final transform matrix. Experimental results demonstrate that the success rate of the proposed method on Apartment and Stairs dataset are 95% and 96%, respectively. The success rate is at least 10% higher than the current commonly-used global and local registration methods. In conclusion, the proposed robust registration method has superior performance in structured scenes with small overlap for applications in different environment.",
keywords = "Robust point cloud registration, coincidence degree, plane elements, plane filtering",
author = "Zhao Zhou and Yaojun Qiao and Aiying Yang",
note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 5th International Symposium on Autonomous Systems, ISAS 2022 ; Conference date: 08-04-2022 Through 10-04-2022",
year = "2022",
doi = "10.1109/ISAS55863.2022.9757263",
language = "English",
series = "2022 5th International Symposium on Autonomous Systems, ISAS 2022",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
booktitle = "2022 5th International Symposium on Autonomous Systems, ISAS 2022",
address = "United States",
}