Several techniques have already been proposed to study the visual perception of the urban environment along a route or a path. Some of them are based on Gibson's concept of motion perspective. Motion perspective refers to the gradual change in the rate of displacement of contour lines in the visual field. Initially developed in the field of aviation psychology, this effect proved to be of paramount importance in the perception of any observer in displacement. In this paper, we propose to develop a computer method to analyze this motion perspective. Our method is based on Teller's spherical metric. It takes advantage of skeletonization, a technique heavily used in image analysis.