A neural network for perceptual grouping, based on architecture of the primary visual cortex, efficient for texture segregation in complex natural images

摘要

A neural network model for perceptual grouping and texture segregation is proposed. This neural net model is based on the architectural and functional properties of the neuronal circuits in the primary visual area of the cerebral cortex. When simulated on the natural and artificial grey-level images, this neural net model is very efficient for segregating textures, even when their local characteristics are highly variable.