Minimal Cases for Computing the Generalized Relative Pose using Affine Correspondences
This work addresses a specific problem in computer vision for multi-camera systems, offering incremental improvements in efficiency and accuracy for motion estimation.
The paper tackles the problem of estimating relative pose for multi-camera systems using affine correspondences, proposing three novel solvers that require fewer correspondences than state-of-the-art methods and achieve superior accuracy on synthetic and real-world KITTI benchmark data.
We propose three novel solvers for estimating the relative pose of a multi-camera system from affine correspondences (ACs). A new constraint is derived interpreting the relationship of ACs and the generalized camera model. Using the constraint, we demonstrate efficient solvers for two types of motions assumed. Considering that the cameras undergo planar motion, we propose a minimal solution using a single AC and a solver with two ACs to overcome the degenerate case. Also, we propose a minimal solution using two ACs with known vertical direction, e.g., from an IMU. Since the proposed methods require significantly fewer correspondences than state-of-the-art algorithms, they can be efficiently used within RANSAC for outlier removal and initial motion estimation. The solvers are tested both on synthetic data and on real-world scenes from the KITTI odometry benchmark. It is shown that the accuracy of the estimated poses is superior to the state-of-the-art techniques.