Observation procedure for a robust visual system

Most visual sensor estimation problems are characterized by nonlinear measurement systems contaminated with noise. In order to get reliable estimation results, the visual system must choose views which permit a maximum of information needed for the task at hand. This is a sensor control problem where an attempt is made to combine measurement parameters with other relevant aspects of the estimation problem to make the best measurement strategy. The purpose of this paper is to describe a suboptimal technique which can be used in conjunction with sensor control to deal with problems that arise in an occlusion case.