A passive stereo calibration technique for visible-thermal, low-resolution imaging in remote sensing applications

When using vision systems that employ multiple spectral bandwidths, such as visible and Long-Wave Infrared (LWIR) bands, the calibration object's appearance can significantly impact the matching of common features in the images. To address this issue, a robust calibration technique is necessary to work with reflections and the degraded appearance of the calibration pattern over time. In this paper, we propose a novel stereo calibration technique that can be used for visible-thermal imaging systems for remote sensing applications (e.g., mounted on Unmanned Aerial Vehicles, UAVs). Proposed method can be used on a generic calibration pattern, and we exploit computer vision algorithms to accurately estimate the intrinsic and extrinsic parameters. The algorithm can mitigate specular reflections and contrast inversion issues that may degrade the calibration accuracy. The calibration technique is designed to be versatile as it operates also in outdoor environments, hence it can be especially useful for UAV applications since on-field re-calibration is often needed. The proposed method has been validated using two different calibration patterns and compared with state-of-the-art techniques.