SIMULATION OF A NEURAL NETWORK FOR ADAPTIVE COLOUR PERCEPTION

The basic network’s unit (neuron) have been designed using a number of logical elements. This neuron has been simulated and tested previously. In the network model three neurons for colour coding are used at each point of the visual field. Three levels of coding are considered. The final level “recognizes” a colour as a point in a three dimensional space of the outputs of the three colour coding neurons. Two of them mainly encode hue and the third one encodes luminosity. Colour contrast is simulated using in addition a set of three neurons, the so-called “adaptors”, one for each type of colour coding neurons. Each adaptor receives excitatory inputs from the same type of colour coding neuron and then inhibits them. In the present simulation of the network’s model, 57 neurons are used for 6 points of the visual field. The results of the simulation have shown that the network “perceives” colours and exhibits adaptive properties of colour contrast and constancy.