A novel approach to attend faces in the crowd through relative visual saliency

Abstract

Visual Saliency plays an important role in attending faces in the crowd. While considering faces in a crowd, humans are inclined towards certain faces due to dominance of some features in those faces like color, texture, intensity, geometry etc. In this paper, we have proposed a novel method to analyze the saliency of faces in a crowd face image based on four parameters namely feature difference (intensity difference), spatial distance, area of each face and camera distance i.e. distance of each face in crowd from the viewer's point. To the best of our knowledge it is confirmed that this zone has not been explored by researchers to its full capacity till now. The experimental results have been found motivating. This method can find its future application in artificial intelligent systems and cognitive models can be established based on this theory which has the capacity to mimic the ability of human vision system.