Abstract
The UAV-based hyperspectral imaging (HSI) has been extensively used in various agriculture applications. The data calibration and pre-processing techniques have been widely studied in the remote sensing research. However, it was ignored that the most important factor affecting the results accuracy is the quality of the spectral data. To acquire such high spectral quality, the bi-directional reflection should be optimally adjusted during the data processing according to radiative transfer equation (RTE) based Hapke’s model. In this study, we quantitatively analyzed the relationship of bi-directional reflectance (BRDF) correction on two crop (Canola and Wheat) varieties in the context of hyperspectral data robustness. The simulated dataset based on the radiative transfer parameters are used to evaluate the influence of source-sensor geometry, light illumination, and scattering on data quality and it’s ability on data correction using proposed spectral similarity measures (SID-SAM score). The RTE-based BRDF optimization method (Hapke model) is proposed for improving the hyperspectral data quality after comparing with the uncorrected datasets. Simulated results shows that the optimization have significantly improved the data quality and increased the signal accuracy. Hence, the bi-directional reflectance correction would definitely have advantages on the hyperspectral data analysis to improve classification results.

This publication has 15 references indexed in Scilit: