Force-maximizing: a tool for image understanding

Abstract

This paper presents the force-maximizing (FMAX) between a particle and an extended object. Landmarks corresponding to desired image properties are located as a result of maximizing the force between particles of predefined masses (i.e., intensities) and positions (i.e., coordinates), and the segment masses of the extended object. FMAX has been tested to determine a vehicle shape configuration and recognize facial expression.