Comparison of two 3D calibration methods for thermal imaging cameras to track bat flight paths near wind turbines

Abstract

Low-resolution 3D cameras intended for use in large outdoor spaces require substantial interaxial distances of several meters between the cameras. This is particularly relevant when studying 3D bat flight paths with thermal cameras near wind turbines to investigate their impact on the animals. Such camera systems need to be 3D calibrated after installation in the field. We have identified two methods that accommodate large interaxial distances: one method (easyWand) relies on landmarks visible in both cameras, while another method we developed (Batflight3D) utilizes the 3D flight path of a drone. In this publication, we aim to provide a representative quantitative comparison of these methods by using the 2D correspondences of the respective drone flight path for both methods, allowing for a direct comparison. The results demonstrate that Batflight3D yields significantly more accurate and reliable results for the aforementioned application but depends on the feasibility of a drone flight and GPS reception.