A Model for Selective Visual Attention Based on Discrete Scale-Spaces

Abstract

A visual selective attention model based on scale-space theory is proposed by employing facts found by neurophysiological experiments and humans’ cognitive phenomena about the vision. The main parts of the proposed model are (a) an early vision model, and (b) a computational model for focus of attention (FOA). Numerical simulations show that (1) the proposed model successively moves FOA to the center of objects, and the result is consistent with ones of humans, (2) the model estimates the appropriate scale (size) for each object, and (3) the model has ability of object detection independent of the sizes and the spatial positions of objects.