Back to search

University of Glasgow

Investigations into applications of photometric stereo and single-pixel imaging

Abstract

dc:description.abstract

Computational image reconstruction is generally an inverse procedure which helps to recover the original information in a scene. Various imaging techniques have been developed to extract certain kinds of information for applications in different fields. The focus of this thesis is to improve two elegant and powerful methods among those approaches, namely, photometric stereo and single-pixel imaging, into a more practical and applicable phase. With the advances in modern imaging technology, 3D information is playing an increasingly significant role in real-world applications, from robotic vision, manufacturing, entertainment, and biology to security. While an immense amount of research has been conducted over the last few decades, the requirement of generating a rapid and accurate estimation of scene depth information with a cost-efficient system remains challenging. In the first work, we developed an inexpensive computational camera system allowing fast 3D reconstruction of objects based on the principle of photometric stereo. By analysing the estimated 3D data of various objects, we noticed good quantitative agreement with the known reference object with a wide viewing angle. With a low-cost accessory, our system provides a simplified reconstruction routine alongside a high efficiency, which extends its portability and capability for practical applications. Single-pixel imaging is an emerging paradigm which utilises spatial correlation of light with a single-pixel detector to form an image. It provides an alternative strategy to conventional imaging techniques which reply on a pixelated sensor for spatial resolution. In the second work, we combined photometric stereo with single-pixel imaging to evolve a new 3D imaging system with an efficient realtime sampling scheme. By utilising a high-speed structured illumination and four single-pixel detectors, multiple images of a scene with different shading profiles were able to be reconstructed with perfect pixel registration for depth estimation, empowering 3D imaging of dynamic scene. A compressive strategy, known as evolutionary compressed sensing, was further employed to improve the frame rate of 3D single-pixel video at an expense of only a modest reduction in image quality. This system represents a step-forward towards real-time 3D single-pixel imaging. By using single-pixel imaging technique, it offers a feasible solution for situations that are costly or constrained with conventional pixelated camera sensor, for instance, near-infrared (NIR) imaging and fluorescence imaging through multimode fibres. However, the signal-to-noise ratio (SNR) scales poorly when increasing the single-pixel imaging resolution. In the last work, we developed a NIR single-pixel imaging system with micro-scanning, an optimisation approach that generates a higher-resolution image while maintaining the SNR of the lower-resolution images where it is derived from. With the use of sunlight and an infrared heat lamp as the illumination sources and a set of NIR bandpass filters, our system indicated a well capability of revealing the water absorption underneath the surfaces of plant leaves and fruits compared to an expensive pixelated InGaAs camera. Additional efforts were devoted to further improve the image quality of a modified single-pixel imaging system that allows visible and NIR dual-band detection simultaneously.

Degree

thesis:*
Level dc:type.qualificationlevel
PhD
Grantor dc:publisher.institution
University of Glasgow
Year dc:date.issued
2017

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Zhang, Yiwei

Subjects

dc:subject × 2

Rights

Language dc:language
en

Chain of custody

source
Harvested from
University of Glasgow
Base URL
theses.gla.ac.uk/cgi/oai2
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Zhang, Yiwei. Investigations into applications of photometric stereo and single-pixel imaging. PhD thesis, University of Glasgow, 2017.