Spectral Reconstruction Using Broadband Inkjet-Printed Optical Filters

Reconstructive spectroscopy offers a promising route toward miniaturization by replacing dispersive optics with computational recovery from encoded measurements, but existing implementations typically rely on complex nanofabrication and highly decorrelated encoding elements, limiting manufacturability and robustness. Here, we demonstrate a reconstructive spectrometer based on inkjet-printed interferometric optical filters featuring a minimalist tri-layer architecture and high optical throughput. Contrary to prevailing design principles, we show that smooth, strongly correlated filter responses can nonetheless enable accurate spectral reconstruction over a 400 nm visible bandwidth. By adaptively selecting optimized subsets of filters using a custom genetic algorithm, the system exploits the high signal-to-noise ratio of correlated encoders while maintaining reconstruction fidelity. In addition, we introduce a calibration and reconstruction framework that requires only minimal a priori knowledge of the measured spectra, increasing the viability of practical deployment of reconstructive spectrometers. The resulting procedure combines scalable fabrication, adaptive encoding, and robust reconstruction, with a strong basis of versatile and manufacturing-ready encoding elements.