Reflectance calibration of UAV-based visible and near-infrared digital images acquired under variant altitude and illumination conditions

Abstract

The understanding of the radiometric response of visible and near-infrared cameras is important for quantitative spectral information obtainment. There is a lack of information regarding the comparison of calibration approaches of near-infrared images of digital cameras, since the Bayer Filter is not usually removed and the obtained NIR image records the infrared radiation in three different spectral bands (red, green and blue). This article aimed at developing reflectance calibration models for UAV-based visible and near-infrared (processed under four different methodologies) digital images acquired under variant altitude and illumination conditions. Reflectance calibration models were developed for red, green, blue and near-infrared spectral bands using UAV-based digital numbers – DNs (acquired in twelve UAV flights under variant altitude and illumination conditions) and laboratory-based reflectance. The variation of DNs among flight conditions represents from 12% to 17%, depending on the spectral band. For the visible bands, all reflectance calibration models presented R2 superior than 0.94. Regarding the near-infrared bands, the grey scale and green channel approaches demonstrated the best results, with R2 between 0.89 and 0.98. The validation of the developed models, assessed by the correlation between reflectance measured in field and the predicted reflectance, revealed strong correlations between the observed and predicted reflectance for the visible bands and for the near-infrared grey scale and green channel approaches, which were found to be more suitable for the calibration of near-infrared digital images into reflectance.