RGB-enhanced quantum dot spectrometers and snapshot spectral imaging with reduced reconstruction error

Quantum dot (QD) filter arrays enable tunable, fabrication-compatible spectral encoding for compact spectrometers and mosaic snapshot spectral imaging. However, the similar high-pass transmittance profiles of QD filters induce strong inter-channel correlation in the sensing matrix, resulting in reduced linear independence among sensing channels and high sensitivity to noise and perturbations. To address this issue, we propose an RGB-enhanced QD joint spectral encoding strategy that combines QD filters with Bayer RGB filters to generate spectral channels with triple the original channel number. For the QD spectrometer, QR decomposition with column pivoting (QR-cp) is used to select a least-correlated subset with the same number of channels as the QD-only system for fair performance evaluation. Simulations show lower reconstruction errors than the baseline QD-only system across multiple noise levels using regularized nonnegative least squares, sparse recovery, and dictionary learning, together with reduced inter-channel correlation and a lower condition number. Moreover, direct use of the all spectral channels without selection (corresponding to triple the original channel number) further improves performance by providing more effective spectral information while maintaining reduced inter-channel correlation. This configuration is also validated experimentally, showing reconstruction error reduction consistent with the simulation results. For QD mosaic snapshot spectral imaging, the same strategy is applied by selecting an equal number of channels for the QD-only and QD&RGB systems via QR-cp to preserve spatial resolution. Extensive evaluations on three public datasets using GAP-net, ADMM-net, and DAUHST under both fixed-noise-power and fixed-SNR settings show that the QD&RGB strategy consistently improves PSNR and structural similarity index measure while reduces spectral angle mapper, confirming its effectiveness for QD mosaic snapshot spectral imaging.