A computational model of object-based selective visual attention mechanism in visual information acquisition

Abstract

A computational model of object-based selective visual attention is introduced. What are the units of selective attention is always the focus of selective attention. Most existing computational model of selective attention is space-based, but more and more neuroscience experiments indicate that it is object-based. For object-based attention, how to define "object" is the biggest difficulty and it is generally accepted that the "object" refers to "perceptual object" which is created by Gestalt rules. However, Gestalt rules are very difficult to be calculated in images. The model presented in this paper is comprised of two independent components. One component is object-based segmentation by using perceptual color cues, the other is the generation of a object-based saliency-map in terms of the inhibition of return mechanism. Lastly, the results of experiments show that the model is valid.