ABSTRACT:CNN models are introduced to analyze and explain the human early vision color perception and some of the visual illusions. CNN model of Horn's algorithm is shown, and a new lightness determination method is introduced. CNN models for the Zöllner, the Café Wall, the Poggendorf, the Brightness, and the Face-Vase illusion are also introduced.

In this paper we present a CNN model for separating colors under different illumination conditions. The color model is based on Land's assumption: the individual monochromatic channels are processed separately. However we use a different channel processing model. The model was evaluated on a Mondrian image.