Bottom-up regions-of-interest in observation of robot hand movement: Comparisons with human experiments

Visual functions are important for robots who engage in cooperative work with other robots. In order to develop an effective visual function for robots, we investigate the features of the human visual scanpath in a scene of robot hand movement. Human regions-of-interest (hROIs) are measured by psychophysical experiments based on eye-movement measurement and different sets of hROIs are compared by using a positional similarity index, S_p, on the basis of scanpath theory. The result reveals consistent loci of the hROIs due to dominant top-down active looking in such a scene. This paper also discusses how bottom-up image processing algorithms (IPAs) are able to predict hROIs. We compare algorithmic regions-of-interest (aROIs), which are generated by the IPAs, with the hROIs. The results suggest that the bottom-up IPAs whose support size is smaller than the size of the fovea have high ability to predict the human regions-of-interest.