Optimal Color Samples for Camera Spectral Sensitivity Estimation

This study investigated the optimal color samples for camera spectral sensitivity estimation. Principal component analysis (PCA) was performed on a public camera spectral sensitivity database. The eigenvectors were extracted and served as the basis functions, and the coefficients were calculated by a classical least square method. The color samples for spectral sensitivity estimation included a multichannel LED device and a reflective color chart. Three sample selection methods were implemented, which were based on minimizing the condition number, maximizing the minimum distance in the chromaticity plane, and using a simulated imaging system, respectively. It was found all the three methods were able to find the optimal color samples. The spectral sensitivity estimation error decreased and then became stable with the increase of the number of color samples.