Matija Milanic

In this study, we introduce a novel hyperspectral imaging approach that leverages variable filament temperature incandescent lamps for active illumination, coupled with multi-channel image acquisition, and provide a comprehensive characterization of the approach. Our methodology simulates the imaging process, encompassing spectral illumination ranging from 400 to 700 nm at varying filament temperatures, multi-channel image capture, and hyperspectral image reconstruction. We present an algorithm for spectrum reconstruction, addressing the inherent challenges of this ill-posed inverse problem. Through a rigorous sensitivity analysis, we assess the impact of various acquisition parameters on the accuracy of reconstructed spectra, including noise levels, temperature steps, filament temperature range, illumination spectral uncertainties, spectral step sizes in reconstructed spectra, and the number of detected spectral channels. Our simulation results demonstrate the successful reconstruction of most spectra, with Root Mean Squared Errors (RMSE) below 5%, reaching as low as 0.1% for specific cases such as black color. Notably, illumination spectrum accuracy emerges as a critical factor influencing reconstruction quality, with flat spectra exhibiting higher accuracy than complex ones. Ultimately, our study establishes the theoretical grounds of this innovative hyperspectral approach and identifies optimal acquisition parameters, setting the stage for future practical implementations.