A benchmark of the implementation of the SIFT in different softwares for 3D reconstruction of terrain surfaces from UAV images

Abstract

In the last decade there has been a significant increase in the use of drones in various applications, among which are the generation of cartographic products. Different software process images captured from these aerial platforms, which due to their weight conditions and vehicle type present greater movements (linear and angular) than those that occur in manned aircraft used in conventional photogrammetry. These conditions are some of the characteristics of the SfM photogrammetry, together with the mathematical concepts that it uses, such as the SIFT transform. In order to compare the effectiveness of the software: Agisoft methashape ®; Pix4D mapper®; Drone2map ®, to implement the SIFT transform, a block with 167 images captured with a UAV TAROT 680 PRO, a low cost camera and supported by 3 GCP was processed to produce a digital surface model and an orthophotograph in each of the three software The results shows that the three software reconstructed the volumetric of the terrain, although qualitatively and quantitatively the best results were obtained with the Agisoft metashape ® software that took longer time and generated the densest 3D point cloud.