Multi-wavelength spectral reconstruction with localized speckle pattern

Abstract

The speckle-based reconstructive spectrometers (RSs) retrieve spectra in a smart and single-shot way, significantly increasing the measurement rate compared to traditional spectrometers with scanning devices. Reports on high-speed multi-wavelength detection in the infrared region are scarce due to limited detector frame rates, unlike in the visible light spectrum, where silicon-based cameras are commonly used for wavemeters. Current RSs commonly rely on full-pixel speckles, but this study demonstrates that localized speckles also convey global spectral information. Experimental results show that an integrating sphere's local speckles can improve spectral measurement speed by 35 times with minimal loss in accuracy compared to full-pixel speckles of a multimode fiber (MMF). The study investigates the influence of the position and size of local speckles on RSs. The optimization criterion for balancing pixel number, measurement speed and reconstruction accuracy is also presented. This research contributes to speedy transmission matrix calibration and detailed observation of agile spectral evolution.