A Novel Approach to Enlighten the Effect of Neighbor Faces during Attending a Face in the Crowd

摘要

While attending the crowd, humans do not look all the faces with same gaze. Human beings have naturally, extraordinary capability to focus their gaze towards a face carrying the 'dominating look' in the crowd in terms of beauty or ugliness. This happens due to relative saliency of that face with respect to their surrounding faces. To model the computer's vision based on human psychology and its cognitive science, enormous researches have been carried out. In order to enhance these researches, this paper proposed a novel method to illustrate how the saliency of a face is varying with its neighboring faces in the crowd. This variation affects our perception and attention towards a face in the crowd. In this method, distribution of visual saliency has been calculated based on intensity values and respective spatial distance of the faces. Experiment has been conceded out on various facial images. The results and accuracy are found to be satisfactory. The evaluation made with the proposed approach which exhibits quite boosting results and it can lead to a future model of intelligent visions of computers.