Receptive Field Neural Network Analysis of Color Constancy and Color Contrast.

Color constancy, or the ability of the visual system to perceive color independently of the ambient illumination, was investigated in the context of a biologically-based neural network. In particular, the role of retinal adaptation and higher level visual operations in mediating color constancy was investigated. The study incorporated properties of individual cells and how they combine to make complex color and spatial operations. The neural network simulations indicate how early visual stages complement each other to compensate and maintain relatively constant color perception under conditions of varying illumination and spatial context -in the image. The network takes advantage of several mechanisms in the human visual system, including retinal adaptation, spectral opponency, and spectrally-specific long-range inhibition. This last stage is a novel mechanism based on cells which have been described in cortical area V4. All stages include non-linear response functions. The model emulates human performance in several psychophysical paradigms designed to test color constancy and color induction. We measured the amount of constancy achieved with both natural and artificial simulated illuminants, using homogeneous gray back-grounds and more complex backgrounds, such as Mondrians. On average, the model performs well or better than the average human color constancy performance index.