Unified study of color sampling.

In this study, we unify some previous research work on the design of optimal color sensors and the reconstruction of color signals from color samples. We show that they are all based on a simple linear vectorspace representation derived from the Karhunen-Loeve (K-L) expansion theorem. We also show how this idea works by two simulation examples. We believe that the theoretical results derived in this study may prove useful both in machine and human color vision research.