Broadband Hybrid Multispectral Sensing for Ripeness Monitoring

Abstract

Spectral sensing has been widely employed in applications ranging from satellite-based remote imaging to biomedicine and precision agriculture. However, broader deployment has been constrained by the complexity and cost of traditional hyperspectral instrumentation. In recent years, efforts have shifted toward the development of compact spectrometers targeting specific spectral regions, often at the expense of broadband analytical capability. In this work, we present the proof-of-concept for an integrated multispectral sensor array for broadband spectral applications, combining Silicon- and GaSb-based photodetectors. Spectral selectivity across the VIS/NIR/SWIR spectral domain (400–2300 nm) is enabled with 19 bandpass dielectric filters in a combined active footprint of 2.02 mm². The sensor assessed in this work was specifically designed to evaluate the optimal harvest time based on weekly spectral measurements from multiple apple cultivars, without the need for destructive chemical analysis. Moreover, we demonstrate a deterministic method that employs a high-scattering region in the NIR as an internal normalization reference, enabling the resolution of temporally evolving spectral signatures associated with carotenoids, anthocyanins, chlorophylls, starch, and moisture content. These results demonstrate that broadband spectral sensors can be deployed at scale in agricultural monitoring with multi-crop validation, enhancing field yield potential and reducing post-harvest losses.