PERFORMANCE ENHANCEMENT OF ENCODING-BASED SINGLE-PIXEL SPECTROMETERS

Abstract

Optical spectrometers have received increasing research and industry attention due to their capability for materials characterization and spectral analysis in different fields, such as chemistry, biology, and medicine. Encoding-based single-pixel spectrometer is a crucial type of spectroscopic system considering its ability for miniaturization. In this thesis, we propose three different single-pixel spectrometers with enhanced performance compared with current encoding-based spectrometers. The first MEMS spectrometer uses compressive sensing theory to design the encoding patterns capable of reconstructing a higher-dimensional signal with fewer measurements, thus enhancing the achievable resolution. The second spectrometer innovatively implements a random Bernoulli patterns to further reduce the required number of measurements for a given dimensional signal. This spectrometer also shows potential in sensing a broadband spectrum. The third spectrometer is a near-infrared spectrometer designed with a wide aperture to improve its optical throughput, thus enhancing its SNR without sacrificing its resolution.