While viewing a crowd, the human vision automatically gets focused towards the most attentive face. This happens when a particular face in the crowd dominates the other faces in terms of beauty, expression, color, shape, size and structure, etc. Human attention towards such faces happens due to its higher visual saliency values. A computer vision system can be modeled based on this aspect of human psychology. Visual saliency of a face in a crowd may vary according to many parameters. In this paper, we propose a novel method to calculate the distribution of visual saliency of faces in the crowd based on their feature difference, spatial distance and size. This method has been tested on various crowd images and inspiring results have been found. Therefore, this can be used to mimic the cognitive behavior of the human vision system to create artificially intelligent computer vision systems.