Facial reconstruction

Search LJMU Research Online

Browse Repository | Browse E-Theses

Integrating 3-D Modelling with Unmanned Aerial Vehicles in Subterranean Environments to aid Archaeological Stratigraphy

Moore, A (2020) Integrating 3-D Modelling with Unmanned Aerial Vehicles in Subterranean Environments to aid Archaeological Stratigraphy. Doctoral thesis, Liverpool John Moores University.

[img] Text
2020MoorePhD.pdf - Published Version
Restricted to Repository staff only until 27 July 2022.

Download (279MB)

Abstract

The aim of this thesis is to integrate Unmanned Aerial Vehicle-based Structure-from-Motion 3-D modelling into subterranean environments to aid the recording of archaeological Strata, overcoming the challenges of acquiring data in a cave. This research is conducted as part of a larger project, the DigiArt Project, № 665066, which deals with the digitisation and dissemination of cultural heritage. The initial experiments are conducted in the Scladina Cave in Belgium. This is an active archaeological site with culturally important historical findings. A 3-D model is created using Structure-from-Motion (SfM) with data captured from a UAV-mounted camera. Although the camera has a wide-angle lens there is only one view captured of the roof and floor of the cave and so they are not properly reconstructed. Other issues arose during this data capture, including the lighting being too warm of a colour-temperature. This proves to be an issue because one of the main aims is to clarify the excavated stratigraphic profiles in the sediment to aid archaeologists. The first experiment from the conclusions of the initial Scladina test is conducted on the image file output from the camera. Many cameras are able to record in a RAW format, however most record in compressed JPEG. The impact this compression has on the model needs to be known so a series of tests are conducted to examine it. The RAW dataset provides the best reconstruction, however good-quality JPEG images are also suitable to be used. Next came a test on the camera types, with the ones being easily attached to the UAV, an action camera, being compared with a mirrorless interchangeable-lens camera. Due to various factors such as better quality optics and a larger, lower noise, sensor, the interchangeable-lens camera produces the best results, however the action camera can still provide good results if the correct settings are used. SfM modelling requires lenses to be as optically perfect as possible to have good reconstructions, however no optical system is perfect. These tests examine lens artefacts along with sensor noise to determine what their impact is on the resulting models. There is an optimum lens sharpness setting, however a lower sensor sensitivity is preferable if balancing the settings. Along with lens artefacts, the software itself needs to be examined and validated. A test is devised to use purely synthetic data so the reconstructed models can be compared against a ground truth model. Two publishers’ software are tested. All reconstructions are acceptable. After the testing is completed, a return trip to Scladina Cave is made. This allows all knowledge gained during the testing to be implemented. The new model of the cave has an increase in coverage of 30% over the old one. The need to clarify the strata is left unfulfilled, so a different approach is taken by post-processing the images. This results in an algorithm which examines the individual colour channel differences in the digital image. The output is a difference image which can be shown on the 3-D model. This new image-processing technique, the testing of the data capture techniques and processing, along with the case-study locations are all the contributions of this thesis.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: 3-D; Modelling; SfM; Structure from Motion; Greece; Archaeology; Stratigraphy
Subjects: C Auxiliary Sciences of History > CC Archaeology
Q Science > QA Mathematics > QA76 Computer software
T Technology > T Technology (General)
Divisions: Engineering
Date Deposited: 29 Jul 2020 16:09
Last Modified: 29 Jul 2020 16:09
DOI or Identification number: 10.24377/LJMU.t.00013351
Supervisors: Shaw, A, Bezombes, F and Cullen, J
URI: https://researchonline.ljmu.ac.uk/id/eprint/13351

Actions (login required)

View Item View Item