Using spherical moments for visual servoing from a special target with unique projection center cameras

This paper focuses on visual servoing from a special target using a vision sensor with an unique projection center. The target is a sphere marked with two points on its surface. Using a spherical projection model, a new minimal set of six features is proposed for this target. Spherical moments are exploited to compute the new set on the image of the sensor. Using the new set, a classical control method is proved to be asymptotically stable even in the presence of modeling errors. Experimental results with a perspective camera and a fish-eye camera validate the proposed theoretical results.