Abstract

This thesis presents the design and test of a compressive imaging (CI) system. In this work, we utilize the compressive sampling (CS) theory to design image-space measurement patterns, which consider images’ sparse energy distribution in the transformed space. The designed patterns are implemented by a Digital Micro-Mirror Device (DMD) to modulate the intensity of images. The modulated image signal is focused by a lens into a single pixel photo detector and is used to reconstruct the original image via convex optimization algorithms. The most significant advantage about the CI system design is the enhancement of detector efficiency, because only significant components in the images’ transformed space, which are important for people’s visual perception, are collected.

In this dissertation, we introduce the hardware realization of three CS measurement methods, namely under-sampling under random basis, variable density sampling in Hadamard space and random sampling in Hadamard space. Images reconstructed from these methods are presented and compared. Also, several application ideas of the proposed CI system are introduced and their experimental realizations are presented. Those ideas include single pixel zoom imaging, single pixel low light imaging, compressive multi-spectral imaging (MSI) and compressive programmable array microscopy (PAM).