Robust Uncertainty-Aware Multiview Triangulation
This work addresses robust 3D point reconstruction from multiple images, which is crucial for computer vision applications like structure-from-motion and SLAM, but it appears incremental as it builds on existing triangulation and uncertainty estimation techniques.
The paper tackles multiview triangulation and uncertainty estimation by proposing an outlier rejection scheme with RANSAC and midpoint method for efficiency, iterative optimization for accuracy, and a learned uncertainty model based on camera count, reprojection error, and parallax angle. It validates the method through extensive evaluation, achieving state-of-the-art efficiency and significant improvements in accuracy and robustness.
We propose a robust and efficient method for multiview triangulation and uncertainty estimation. Our contribution is threefold: First, we propose an outlier rejection scheme using two-view RANSAC with the midpoint method. By prescreening the two-view samples prior to triangulation, we achieve the state-of-the-art efficiency. Second, we compare different local optimization methods for refining the initial solution and the inlier set. With an iterative update of the inlier set, we show that the optimization provides significant improvement in accuracy and robustness. Third, we model the uncertainty of a triangulated point as a function of three factors: the number of cameras, the mean reprojection error and the maximum parallax angle. Learning this model allows us to quickly interpolate the uncertainty at test time. We validate our method through an extensive evaluation.