Abstract

Traditional optical imaging systems have constrained angular and spatial resolution, depth of field, field of view, tolerance to aberrations and environmental conditions, and other image quality limitations. Computational imaging provided an opportunity to create new functionality and improve the performance of imaging systems by encoding the information optically and decoding it computationally. The design of a computational imaging system balances hardware costs and the accuracy and complexity of the algorithms. In this thesis, two computational imaging systems are presented: Randomized Aperture Imaging and Laser Suppression Imaging. The former system increases the angular resolution of telescopes by replacing a continuous primary mirror with an array of light-weight small mirror elements, which potentially allows telescopes to have very large diameter at a reduced cost. The latter imaging system protects camera sensors from laser effects such as dazzle by use of a phase coded pupil plane mask. Machine learning and deep learning based algorithms were investigated to restore high-fidelity images from the coded acquisitions. The proposed imaging systems are verified by experiment and numerical modeling, and improved performances are demonstrated in comparison with the state-of-the-art.

Publication Date

8-5-2022

Document Type

Dissertation

Student Type

Graduate

Degree Name

Imaging Science (Ph.D.)

Department, Program, or Center

Chester F. Carlson Center for Imaging Science (COS)

Advisor

Grover Swartzlander

Advisor/Committee Member

Linwei Wang

Advisor/Committee Member

Zoran Ninkov

Campus

RIT – Main Campus

Share

COinS