Single-pixel, single-photon three-dimensional imaging

QC Physics

Musarra, Gabriella ; Faccio, Daniele

University:University of Glasgow

Department:School of Physics and Astronomy

关键词: 3D imaging, single-pixel imaging, neural networks, computational imaging.;   

Others  :  http://theses.gla.ac.uk/81446/1/2020MusarraPhD.pdf

来源: University of Glasgow

【 摘 要 】

The 3D recovery of a scene is a crucial task with many real-life applications such as self-driving vehicles, X-ray tomography and virtual reality. The recent development of time-resolving detectors sensible to single photons allowed the recovery of the 3D information at high frame rate with unprecedented capabilities. Combined with a timing system, single-photon sensitive detectorsallow the 3D image recovery by measuring the Time-of-Flight (ToF) of the photons scattered back by the scene with a millimetre depth resolution. Current ToF 3D imaging techniques rely on scanning detection systems or multi-pixel sensor. Here, we discuss an approach to simplify the hardware complexity of the current 3D imaging ToF techniques using a single-pixel, single-photon sensitive detector and computational imaging algorithms. The 3D imaging approaches discussed in this thesis do not require mechanical movingparts as in standard Lidar systems. The single-pixel detector allows to reduce the pixel complexity to a single unit and offers several advantages in terms of size, flexibility, wavelength range and cost. The experimental results demonstrate the 3D image recovery of hidden scenes with a subsecondacquisition time, allowing also non-line-of-sight scenes 3D recovery in real-time. We also introduce the concept of intelligent Lidar, a 3D imaging paradigm based uniquely on the temporal trace of the return photons and a data-driven 3D retrieval algorithm.