Globally Optimal Relative Pose Estimation with Gravity Prior
This work provides a more efficient and accurate method for relative pose estimation for systems equipped with IMUs, such as smartphones, cars, and UAVs, by simplifying the problem with a gravity prior.
This paper addresses relative pose estimation by leveraging gravity vector measurements from IMUs to align camera y-axes, thereby reducing the relative orientation to a single degree-of-freedom. The authors propose a novel globally optimal solver that minimizes algebraic error and a fast solver using a first-order approximation, demonstrating their effectiveness on four real-world datasets with approximately 50,000 image pairs.
Smartphones, tablets and camera systems used, e.g., in cars and UAVs, are typically equipped with IMUs (inertial measurement units) that can measure the gravity vector accurately. Using this additional information, the $y$-axes of the cameras can be aligned, reducing their relative orientation to a single degree-of-freedom. With this assumption, we propose a novel globally optimal solver, minimizing the algebraic error in the least-squares sense, to estimate the relative pose in the over-determined case. Based on the epipolar constraint, we convert the optimization problem into solving two polynomials with only two unknowns. Also, a fast solver is proposed using the first-order approximation of the rotation. The proposed solvers are compared with the state-of-the-art ones on four real-world datasets with approx. 50000 image pairs in total. Moreover, we collected a dataset, by a smartphone, consisting of 10933 image pairs, gravity directions, and ground truth 3D reconstructions.