Thesis

Smart illumination photometric stereo based 3D imaging system using LED technology

Downloadable Content

Download PDF
Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2022
Thesis identifier
  • T16378
Person Identifier (Local)
  • 201753197
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • The high-modulation rate of light-emitting diodes combined with a new modulation scheme called Manchester Encoded Binary Frequency Division Multiple Access enables the self-synchronisation of a set of white light-emitting diodes between themselves and a camera. Based on this smart illumination system, a new synchronisation-free photometric stereo imaging system is presented to achieve high-resolution 3D shape reconstruction in order to enhance indoor video surveillance systems. This thesis first demonstrates the experimental proof-of-concept of the top-down illumination photometric stereo imaging system using off-the-shelf equipment such as commercially available white light-emitting diodes and a smartphone. A depth resolution of 3 mm for an object imaged at a distance of 42 cm and dimensions of 48 mm is reported. Dynamic imaging application of the experimental setup achieved a full 3D reconstruction of an ellipsoid in motion at a video rate of 25 fps with an error ranging between 4 mm and 11 mm at a similar distance. A hybrid imaging system based on time-of-flight and photometric stereo imaging is also reported in this work. The former can achieve depth accuracy in discontinuous scenes and the latter can reconstruct surfaces of objects with fine depth details and high spatial resolution. The experimental proof-of-principal shows a root mean square error ranging between 4% and 5% for an object auto-selected from a scene imaged at a distance of 50 cm to 70 cm. Finally, the generation of a bespoke synthetic dataset for top-down illumination photometric stereo imaging is achieved. This dataset aims to constitute the building blocks of a potential convolutional neural network which would decrease the computational time of the global image processing in order to reach real-time imaging applications.
Advisor / supervisor
  • Dawson, Martin D.
Resource Type
DOI

Relations

Items

Thumbnail Title Date Uploaded Visibility Actions
file details PDF of thesis T16378 2022-11-07 Public Download
file details File 2022-10-25 Private