Date of Award:

12-2025

Document Type:

Thesis

Degree Name:

Master of Science (MS)

Department:

Electrical and Computer Engineering

Committee Chair(s)

Scott E. Budge

Committee

Scott E. Budge

Committee

Jonathan Phillips

Committee

Jacob Gunther

Abstract

Drones equipped with laser scanners (LiDAR) and cameras capture detailed 3D scenes. Combining the laser points with photos builds realistic, photo-textured 3D maps used in surveying, agriculture, and infrastructure planning. Conventional methods to create these maps often misrepresent inward shapes such as doorways, overhangs, and outcrops. These shapes are misrepresented because drones mainly view top surfaces and do not collect significant data on vertical or hidden areas.

This study introduces a surface-reconstruction method that groups LiDAR points into clusters, reconstructs smooth surfaces for each cluster, and uses information from recorded camera poses to stitch clusters together. Tests on real drone flights with synchronized LiDAR, imagery, and navigation data show that the method better preserves concave features, mitigates errors from sparse sampling, and improves the completeness and accuracy of the resulting 3D maps.

Checksum

fe51d4394231a625b90a966d9384bbb3

Share

COinS