Efficient rotation estimation for 3D registration and global localization in structured point clouds

作者：

Highlights：

• A direction angle of normal vectors is defined.

• A rotation estimation algorithm is proposed using the direction angles.

• Our algorithm is applied to 3D registration and indoor robot global localization.

• Our algorithm is more robust to noise, missing data, small overlaps and outliers.

• Our algorithm achieves a higher accuracy and computational efficiency.

摘要

•A direction angle of normal vectors is defined.•A rotation estimation algorithm is proposed using the direction angles.•Our algorithm is applied to 3D registration and indoor robot global localization.•Our algorithm is more robust to noise, missing data, small overlaps and outliers.•Our algorithm achieves a higher accuracy and computational efficiency.

论文关键词：Structured point clouds,Small overlaps,Registration,Global localization,ICP,Direction angle

论文评审过程：Received 23 June 2016, Revised 12 July 2017, Accepted 9 September 2017, Available online 20 September 2017, Version of Record 4 October 2017.

论文官网地址：https://doi.org/10.1016/j.imavis.2017.09.003