Comparison of Radiometric Correction Methods for Errors Caused by Large Variance in Sun Irradiance

In recent years, UAVs (Unmanned Aerial Vehicles) equipped with sunlight sensors have become increasingly available and easy to operate. However, it is difficult to perform accurate sensing with them, especially in cloudy weather. Moreover, the error in NDVI (Normalized Difference Vegetation Index) images caused by a large variance in sun irradiance leads to unnatural seam lines in the NDVI images generated by Pix4Dmapper. In this study, we suggest two methods of radiometric correction for these artifacts: Method A, which scales the irradiance values, and Method B, which converts the pseudo-reflectance. According to a visual assessment, the method of converting pseudo-reflectance removes the seam lines better than scaling the irradiance values. By contrast, the accuracy of the NDVI images is better preserved by scaling the irradiance values. Although it is best to perform sensing with UAVs under clear conditions, in situations in which time constraints make this infeasible, the findings of this study may serve as a guide.