Unmanned Aerial Vehicle Path Planning and Image Processing for Orthoimagery and Digital Surface Model Generation

Abstract

Due to their relatively cheap costs and ability to fly at low altitudes above ground, micro unmanned aerial vehicles are ideal platforms for performing photogrammetric missions above archaeological sites. Advances in image matching and 3D point-cloud generation from 2D images have allowed easier generation of digital surface models and orthophotographs from images captured by an unmanned aerial vehicle equipped with a high-resolution camera. These digital surface models and orthophotographs are much higher resolution and generated in a timelier manner than those from traditional methods, such as satellites, kites, balloons and total stations. However, current unmanned aerial vehicle systems require a high level of technical knowledge or pilot ability to perform photogrammetric tasks. This thesis seeks to make the entire process of generating digital surface models and orthophotographs simpler, from capturing the images to processing them, by presenting a new path planning algorithm that optimizes over various parameters. Simulations showed that choosing a path, which minimizes the number of flight lines across the site being photographed, by accounting for geometric properties of the site, performs the best, even in the presence of wind. Furthermore, various parameters were explored using Agisoft’s Photoscan to generate digital surface models and orthophotographs from images captured by an unmanned aerial vehicle flown manually over an archaeological site in Peru. Those experiments with Photoscan revealed several, subjective image quality conditions for guaranteeing better image matching and confirmed that a back-and-forth path produces the best matching and quality of digital surface models and orthophotographs.