A Wide-Baseline Binocular Vision Extrinsic Calibration Method Assisted by Unmanned Aerial Vehicle (UAV)

Accurate extrinsic calibration of a wide-baseline binocular vision system is essential for reconstructing precise three-dimensional positions of aerial swarm. However, traditional extrinsic calibration methods face challenges due to the scarcity of texture features in the sky. To address this issue, we propose a wide-baseline binocular vision extrinsic calibration method assisted by an unmanned aerial vehicle (UAV). First, the UAV carries two rigidly connected balls to provide matched points for system calibration. Second, the three-dimensional positions of the two balls are reconstructed through triangulation, and the extrinsic parameters are optimized to minimize the reconstruction error in the calculated distance between the balls. Experimental results demonstrate that our method can effectively enhance the accuracy of extrinsic calibration for wide-baseline binocular vision systems in both simulated and physical environments.