A Geometric Estimation Technique Based on Adaptive M-Estimators: Algorithm and Applications

Robust fitting is a basic technique and has been widely applied in photogrammetry remote sensing, such as geometric correction. As known, typical robust estimators (include M-estimators, S-estimators, MM-estimators, etc.) often fail when outlier rate higher than 50%, even if the outliers are uniformly distributed. In this article, we propose simple yet effective estimators, called adaptive M-estimators (AM-estimators). They still under 80% of outliers. The proposed AM-estimators very important supplements M-estimators. Different from use varying parameter (shape-control parameter) instead original constant weight function. shape-control decreases along with iterations iteratively reweighted least squares, namely, optimized coarse-to-fine manner. We adapt into classical sensing tasks, including mismatch removal, camera orientation (or perspective-n-point), point set registration to demonstrate their powers. Extensive synthetic real experiments show that superior RANSAC-type methods. source code will be publicly available at https://ljy-rs.github.io/web.