Thermal camera geometric self-calibration supported by RTK measurements

Thermal InfraRed (TIR) imaging sensors are already successfully used in several agriculture applications, for instance providing valuable information for assessing plant health, scheduling crop irrigation regimes, and, more in general, to support site-specific decision strategies based on monitoring crop stress and phenotyping. Once mounted on Unmanned Aerial Systems (UASs), they can be used for plant health monitoring purposes, collecting geospatial data at both high spatial and temporal resolutions. High resolution spatio-temporal monitoring requires the production of highly accurate data, which can only be ensured when the TIR sensors are well calibrated. This paper compares the performance of several techniques for the production of photogrammetric reconstructions, focusing in particular on the effect of GNSS measurements on the geometric self-camera calibration results, and on the consequent repercussions on the obtained 3D models.