This package implements relative pose estimation for multi-camera systems [1] and monocular cameras [2] from point correspondences with scale ratio. The solver for monocular cameras is a re-implementation of [2] using the automatic solver generator of Larsson et al [3].
Source codes and Matlab mex files with demo code are provided in the package. The core solvers are written by C++. Matlab mex files are compiled using Ubuntu 16.04 + Matlab R2019a. Run test_solvers_sift.m in folder "test" to start.
[1] Banglei Guan, and Ji Zhao. Relative Pose Estimation for Multi-Camera Systems from Point Correspondences with Scale Ratio. ACM Multimedia, 2022.
[2] Stephan Liwicki and Christopher Zach. Scale Exploiting Minimal Solvers for Relative Pose with Calibrated Cameras. British Machine Vision Conference (BMVC), 2017.
[3] Viktor Larsson, Kalle Astrom, and Magnus Oskarsson. Efficient Solvers for Minimal Problems by Syzygy-based Reduction. IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR), 2017.
If you use this package in an academic work, please cite:
@inproceedings{guan2022relative,
title={Relative Pose Estimation for Multi-Camera Systems from Point Correspondences with Scale Ratio},
author={Guan, Banglei and Zhao, Ji},
booktitle={ACM Multimedia},
year={2022},
pages={5036--5044}
}
Four minimal solvers for the relative pose estimation of multi-camera systems using three point correspondences with scale ratio. Returns a maximum of 4 solutions, 8 solutions, or 16 solutions.
-
Solvers:
solver_gcam_sift_3inter.mexa64solver_gcam_sift_3intra.mexa64solver_gcam_sift_2inter_1intra.mexa64solver_gcam_sift_1inter_2intra.mexa64 -
API:
[qt_sols, R_sols] = solver_gcam_sift(data, match_type);
Compiled files using Ubuntu 16.04 + Matlab R2019a are provided. You can run the package in Matlab.
test_solvers_sift.m is the demo which shows how to call the APIs.