Guaranteed Outlier Removal for Rotation Search

摘要: Rotation search has become a core routine for solving many computer vision problems. The aim is to rotationally align two input point sets with correspondences. Recently, there significant interest in developing globally optimal rotation algorithms. A notable weakness of global algorithms, however, their relatively high computational cost, especially on large problem sizes and data proportion outliers. In this paper, we propose novel outlier removal technique search. Our method guarantees that any correspondence it discards as an does not exist the inlier set original data. Based simple geometric operations, our algorithm deterministic fast. Used preprocessor prune portion outliers from data, enables substantial speed-up algorithms without compromising optimality. We demonstrate efficacy various synthetic real experiments.