The human retina encodes information about images through the responses of three classes of photoreceptor, often referred to as the L, M, and S cones. These photoreceptors are arranged in three interleaved mosaics; at any one retinal location only a single cone type samples the retinal image. To create our percept of a continuous colored world, the visual system must reconstruct the responses of the missing two cone types at each retinal location. The algorithm that performs this reconstruction works very well—we rarely perceive artifacts that arise from the interleaved sampling arrangement.Most CCD based color cameras employ the same interleaved sampling architecture as the human retina. Yet for CCD cameras, color artifacts are quite common near sharp luminance boundaries. CCD cameras are more susceptible to these artifacts because their reconstruction algorithms are not as successful as the one employed by the human visual system.This talk will begin by reviewing basic research designed to elucidate reconstruction by the human visual system. We will then show how ideas that emerged from the basic research have led to a new algorithm for processing images acquired with CCD cameras.
David H. Brainard, Doron Sherman, "Reconstructing Images from Trichromatic Samples: From Basic Research to Practical Applications" in Proc. IS&T 3rd Color and Imaging Conf., 1995, pp 4 - 10, https://doi.org/10.2352/CIC.1995.3.1.art00002