Home

Awesome

README

This package implements relative pose estimation for multi-camera systems from six point correspondences, including a generic minimal solver for the generalized camera (solver_trans_generic_6pc) and two minimal solvers for popular configurations of two-camera rigs (solver_trans_inter_6pc, solver_trans_intra_6pc).

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_solver_pc.m in folder "test".

Reference

[1] Banglei Guan, Ji Zhao, and Laurent Kneip. Six-Point Method for Multi-Camera Systems with Reduced Solution Space. European Conference on Computer Vision, 2024.

If you use this package in an academic work, please cite:

@inproceedings{guan2024six,
  title={Six-Point Method for Multi-Camera Systems with Reduced Solution Space},
  author={Guan, Banglei and Zhao, Ji and Kneip, Laurent},
  booktitle={European Conference on Computer Vision},
  year={2024}
 }

The code was written by Ji Zhao. The initial version of this code was developed in 2020 and is available at GitLab

solver_trans_generic_6pc

A generic minimal solver for the relative pose estimation of generalized camera using six point correspondences. Returns a maximum of 64 solutions.

solver_trans_inter_6pc

A minimal solver for the relative pose estimation of two-camera rigs using six inter-camera point correspondences. Returns a maximum of 48 solutions or 56 solutions.

solver_trans_intra_6pc

A minimal solver for the relative pose estimation of two-camera rigs using six intra-camera point correspondences. Returns a maximum of 48 solutions.

Run

Compiled files using Ubuntu 16.04 + Matlab R2019a are provided. You can run the package in Matlab.

test_solver_pc.m is the demo which shows how to call the APIs.