Abstract

Spectral and polarization information are crucial for characterizing the composition and surface morphology of various materials. However, traditional spectral and polarization detection systems are hampered by bulky, single-function optical components and complex configurations, hindering the portable, low-cost, and multifunctional applications in compact, field-deployable devices.

In this study, we introduce a miniaturized near-infrared (NIR) spectro-polarimetric detection device utilizing a group of meta-spectro-polarimeters (MSPs), which support quasi-bound states in the continuum (Q-BIC) under specific incident polarization, significantly reducing the spatial footprint and improving the design scalability.

By precisely adjusting the high quality resonances and polarization sensitive radiation behavior of MSPs, we not only extend the number of spectral waveband to 20 that greatly surpassing traditional division-of-amplitude or division-of-time schemes, but also ensure an average spectral resolution power (SRP, λ/Δλ) of 71.03.

We achieve high accuracy multidimensional spectro-polarization detection, resulting in the angle of linear polarization (AoLP) with an average error of 4.67° and precise spectrum reconstruction in a machine-learning free way, avoiding the huge computational power and energy during the training process.

Additionally, we demonstrate the high spectro-polarimetric imaging performance of our system in different experimental scene. Such multidimensional detection device holds great promise for adoption in fields like aerospace, chemical detection, machine vision, and so on.