AI for time-resolved imaging: from fluorescence lifetime to single-pixel time of flight

Kapitány, Valentin (2023) AI for time-resolved imaging: from fluorescence lifetime to single-pixel time of flight. PhD thesis, University of Glasgow.

Abstract

Time-resolved imaging is a field of optics which measures the arrival time of light on the camera. This thesis looks at two time-resolved imaging modalities: fluorescence lifetime imaging and time-of-flight measurement for depth imaging and ranging. Both of these applications require temporal accuracy on the order of pico- or nanosecond (10−12 − 10−9s) scales.

This demands special camera technology and optics that can sample light-intensity extremely quickly, much faster than an ordinary video camera. However, such detectors can be very expensive compared to regular cameras while offering lower image quality. Further, information of interest is often hidden (encoded) in the raw temporal data. Therefore, computational imaging algorithms are used to enhance, analyse and extract information from time-resolved images.

"A picture is worth a thousand words". This describes a fundamental blessing and curse of image analysis: images contain extreme amounts of data. Consequently, it is very difficult to design algorithms that encompass all the possible pixel permutations and combinations that can encode this information. Fortunately, the rise of AI and machine learning (ML) allow us to instead create algorithms in a data-driven way. This thesis demonstrates the application of ML to time-resolved imaging tasks, ranging from parameter estimation in noisy data and decoding of overlapping information, through super-resolution, to inferring 3D information from 1D (temporal) data.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science; Q Science > QC Physics
• Colleges/Schools: College of Science and Engineering > School of Physics and Astronomy
• Supervisor's Name: Faccio, Professor Daniele and Murray-Smith, Professor Roderick
• Date of Award: 2023
• Depositing User: Theses Team
• Unique ID: glathesis:2023-83708
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 05 Jul 2023 14:19
• Last Modified: 05 Jul 2023 14:19
• Thesis DOI: 10.5525/gla.thesis.83708
• URI: https://theses.gla.ac.uk/id/eprint/83708

Related URLs

• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Data DOI
• Data DOI
• Data DOI