DEM modeling using RGB-based vegetation indices from UAV images

Abstract

Traditional NDVI techniques require NIR images from multispectral cameras in order to identify vegetation. Research indicates that RGB images from UAV platforms can provide a cost-efficient and near-real time survey with high temporal and spatial resolution. In this study, only RGB images taken with a 20MP camera mounted on a UAV glider were used to conduct a ground survey and generate a Digital Elevation Model (DEM). Over 7,000 UAV images with less than 5cm ground resolution were used in order to survey a 5km2 in the Alassa region in Cyprus in order to produce a DEM. The area was geo-referenced using ground control points. Due to extensive vegetation coverage, a RGB-based vegetation indice was used to mask the vegetation and produce a DEM using interpolation techniques. This study highlights a cost-effective technique to survey and model large areas with vegetation coverage.