Certifiable relative pose estimation

In this paper we present the first fast optimality certifier for non-minimal version of Relative Pose problem calibrated cameras from epipolar constraints. The proposed is based on Lagrangian duality and relies a novel closed-form expression dual points. We also leverage an efficient solver that performs local optimization manifold original problem's non-convex domain. solution then checked via our certifier. extensive conducted experiments demonstrate that, despite its simplicity, certifiable excellently synthetic data, repeatedly attaining (certified posteriori) optimal shows satisfactory performance real data.