| Author: | Huang, Weijie |
| Title: | Deep-learning-based ghost imaging |
| Advisors: | Chen, Wen (EEE) |
| Degree: | M.Sc. |
| Year: | 2024 |
| Subject: | Quantum optics Image processing Deep learning (Machine learning) Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Electrical and Electronic Engineering |
| Pages: | iii, 50 pages : color illustrations |
| Language: | English |
| Abstract: | Traditional ghost imaging (GI), also known as quantum imaging or single-pixel imaging, is an imaging technique that recovers an image of an object by utilizing information from two optical detectors simultaneously: a single-pixel (bucket) detector that views the object, a traditional detector that does not sense the object. Although the GI system has superior advantages over the traditional optical imaging system, GI has a severe problem: the imaging speed of GI is not fast enough to maintain image quality. Most research focuses on improving the data acquisition or processing method, neither resolves the imaging speed problem. In this dissertation, deep learning (DL) is employed as a novel technique to improve the reconstructed image quality and reduce the imaging time. Simulations of speckle patterns and the convolutional neural network (CNN) are done to investigate the model. More specifically, optimizing the parameter of the CNN to fit the characteristics of DGI enables us to reconstruct higher-quality images. The CNN model is trained at both the high and low sampling rates and dataset sizes, and the effect of adjusting the hyperparameter of the CNN model is discussed at 100% sampling rate. By applying DL for differential ghost imaging (DGI) reconstruction, speckle noise in the recovered images is reduced, and the object to be detected can be imaged quickly in high quality. Compared to other methods that use compressed sensing and basic correlation methods, the modified CNN can learn from a massive dataset of images to alleviate the noise and artifacts, leading to faster imaging speed and more accurate reconstructed images. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 8313.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 4.36 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13905