Spatio-temporal fusion of visual attention model

The motivation behind a spatio-temporal visual saliency model is to extract salient information from two distinct pathways: static (intensity) and dynamic (motion). Consequently, the information from these pathways is combined to get the final visual saliency map. Since the response of the pathways is different, the step of combination of the maps is important. As a consequence, we study six recent fusion techniques against two video databases using human eye positions from an eye tracker. A criterion is used to evaluate and underline the significance of these fusion methods.