In the evaluation and optimal design of digital camera spectral sensitivity (SS), it is useful to have a metric of goodness that agrees closely with the perceived color accuracy of human visual system. In this paper, Vora et al's μ-factor based on the orthonormal color matching function space and a universal measure of goodness (UMG) based on CIE L*a*b* space with noise consideration were introduced to evaluate and optimize the camera spectral sensitivities. General physical constraints on spectral sensitivities were discussed in the paper and three feasible approaches to the optimal design of camera spectral sensitivities were proposed, compared and verified with simulated experiments.