In this study, we present a comparison of popular methods for the interpolation of irregular spatial data in order to determine the applicability of each algorithm for hyperspectral reflectance estimation. The algorithms were benchmarked against a very high-resolution orthoimage from an RGB camera and medium-resolution satellite imagery from Sentinel-2A. We tested five interpolation algorithms: Triangulated Irregular Network (TIN), Inverse Distance Weighting (IDW), Ordinary Kriging (OK), Universal Kriging (UK), and Kriging with external drift (KED) for transformation of spectrometer reflectance footprint into reflectance images. Next we tested four downsampling methods: Box filtering, Gaussian, Catmull-Rom (Catrom) and Lanczos to compare the interpolated reflectance images with Sentinel-2 reflectance. Overall the best interpolation was achieved with the KED algorithm, and the lowest errors were achieved with the Gaussian downsampling method.
Authors
Additional information
- DOI
- Digital Object Identifier link open in new tab 10.1109/igarss52108.2023.10282982
- Category
- Aktywność konferencyjna
- Type
- publikacja w wydawnictwie zbiorowym recenzowanym (także w materiałach konferencyjnych)
- Language
- angielski
- Publication year
- 2023
