An Accurate Outlier Rejection Network With Higher Generalization Ability for Point Cloud Registration

Shiyi Guo, Fulin Tang, Bingxi Liu, Yujie Fu, Yihong Wu

Abstract: Feature-based point cloud registration algorithms have gained more attention recently for their high robustness. Outlier rejection is a key step of such algorithms. With the development of deep learning, some of the learning-based outlier rejection methods have been proposed and implemented in various scenes. However, generalization ability and accuracy of the existing methods in complex scenes still need to be improved. In this letter, we construct a neural network for removing outlier correspondences. Particularly, we propose a novel seed selection method based on feature consistency (FC) and a new loss function based on second order feature consistency (FC <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{2}$</tex-math></inline-formula> ). Experimental results on various datasets show the proposed network achieves better accuracy and stronger generalization ability than the state-of-the-art learning-based algorithms.