Rapid 3D Scene Reconstruction from Kite-Based Aerial Imagery Using Open Source Structure from Motion

Abstract

There has been a significant increase over the past decade in worldwide demand for rapid and efficient capturing of 3D geospatial data. Until recently the technology and training essential to acquire 3D geospatial data was severely limited to government agencies or large corporations. Increasingly affordable tools and methods have given 3D geospatial capabilities to the public, which can be observed within numerous fields of research and industry, such as: military operations, emergency response, humanitarian aid, municipal upkeep, topographic surveying, and cultural heritage. With applicable 3D data the ability to visualize surroundings, perform simulations, conduct crucial research, and gain invaluable knowledge becomes possible. This thesis focuses on developing an accessible, affordable workflow for rapidly creating 3D scenes constructed from captured image series by an aerial platform. By using open source software and simple to operate aerial vehicles such as kites, this workflow is able to be used in nearly any application while requiring minimal training and funding.