Automatic Design of High-Sensitivity Color Filter Arrays With Panchromatic Pixels

摘要: In most of existing digital cameras, color images have to be reconstructed from raw images, which only one sensed at each pixel, as their imaging sensors are covered by filter arrays (CFAs). At a CFA usually allows portion the light spectrum pass through, and thereby reduces sensitivity pixels. To address this issue, previous works explored adding panchromatic pixels into CFAs. However, almost all methods assign empirically , making designed CFAs prone aliasing artifacts. paper, based on mathematical model we propose fully automatic approach designing high-sensitivity using By frequency structure representation CFAs, formulate design continuous multi-objective optimization problem, where robustness artifacts percentage simultaneously maximized. We analyze characteristics our new formulation. According analysis, develop method candidates, can produce that reach desired Then for candidate, optimize parameters obtain final CFA, is an appropriately balanced solution problem. two procedures constrained problems solve them alternating direction method. Extensive experiments confirm advantage proposed in both low-light normal-light conditions.