Superiority of optimal broadband filter sets under lower noise levels in multispectral color imaging

Abstract

AbstractOptimizing multispectral cameras with broadband or narrowband optical filters is necessary for accurate color reproduction and spectral reconstruction. However, optimal broadband filter sets can provide better performance than narrowband filter sets under certain circumstances; previous studies have not addressed this. Herein, after selecting optimal filter sets by using the multi‐objective optimization method, the optimal broadband and narrowband filter sets were systematically compared. Ten different noise variation levels and five distinct spectral channels were employed to process the imaging simulations. Moreover, typical commercial absorption glass filters were used as the broadband filters, and the transmittance curves of the narrow filters were generated by Gaussian functions with varying parameters rather than specific physical modalities. The results indicated that all the selected optimal broadband filter sets exhibited a better performance at lower noise levels, and all the optimal narrowband filter sets had comparatively stronger anti‐noise features, although the optimal filter sets had a better performance than the suboptimal filter sets for both broadband or narrowband filter sets. The dependence of the two different types of the filters on noise levels for optimal performance is significant for designing multispectral cameras, especially for broadband multispectral imaging sensors during high‐quality spectral acquisition.