Recent technological innovations, eg detection and acquisition hardware, have pushed sparse-photon imaging to the fore in a variety of applications including 3D Lidar imaging and microscopy. 3D Lidar imaging consists in sending laser pulses to a target and capturing the returned photons after reflection from the target. Recent advances in single-photon detectors allowed the use of such systems to acquire 3D images in low photon regime (few received photons) due for example to long-range km imaging or fast imaging, which constitute important challenges for automotive Lidar and sensing for autonomous vehicles. Despite recent advances, current systems can still be optimized regarding the task to be achieved such as parameters estimation, classification, etc.
In this project, we will consider optimizing both the acquisition and processing of the single-photon data to achieve better performance in estimating the parameters of interest (e.g., depth and reflectivity information from a target in 3D imaging). The PhD candidate will improve the acquisition by studying compressive and/or adaptive sampling strategy to select and acquire informative data regarding the estimation task, which will lead to a non-uniform sampling of the data. Due to the challenging acquisition conditions used in this project, we will need to restore the acquired images by considering advanced processing methods. A focus will be on the combination of statistical Bayesian models with state-of-the-art deep learning algorithms to solve this challenging inverse problem (e.g., plug-and-play, and unrolling approaches). The candidate will also investigate the exploitation of complementary information from different imaging sensors/modalities to improve performance (e.g., use of high-resolution RGB image to fill sparse depth images as required in autonomous driving). The developed methods will be validated on several sensing/imaging applications including Lidar depth imaging, microscopy imaging, hyperspectral imaging and satellite altimetry.
Through the project, the PhD candidate will learn state-of-the-art approaches regarding Bayesian modelling, deep learning architectures, non-local filtering, graph-based approaches, and optimization algorithms. The project will be achieved in collaboration with industrial partners and system design teams in HWU which will provide additional real data.
More information regarding the group can be accessed in: https://sites.google.com/site/abderrahimhalimi/home
Software Needs and Skills:
Statistical signal and Image processing, Bayesian methods, deep learning.
Matlab, Python, C/C++.
How to Apply
1. Important Information before you Apply
When applying through the Heriot-Watt on-line system please ensure you provide the following information:
(a) in ‘Study Option’
You will need to select ‘Edinburgh’ and ‘Postgraduate Research’. ‘Programme’ presents you with a drop-down menu. Choose Chemistry PhD, Physics PhD, Chemical Engineering PhD, Mechanical Engineering PhD, Bio-science & Bio-Engineering PhD or Electrical PhD as appropriate and select September 2022 for study option (this can be updated at a later date if required)
(b) in ‘Research Project Information’
You will be provided with a free text box for details of your research project. Enter Title and Reference number of the project for which you are applying and also enter the potential supervisor’s name.
This information will greatly assist us in tracking your application.
Please note that once you have submitted your application, it will not be considered until you have uploaded your CV and transcripts.