Evaluation of the Robustness of Visual Behaviors through Performance Characterization

Abstract. The robustness of visual behaviors in an active vision system depends both on the vision algorithms and control structure. Therefore improvements in robustness depend strongly on the evaluation of the performance of the several algorithms. Performance analysis can be done within the framework of control theory since notions such as stability and controllability can contribute to a better understanding of the algorithms and architectures weaknesses. In this paper we discuss the generation of reference trajectories and we characterize the performance of smooth pursuit. This system is used to track binocularly targets in a surveillance environment. One of the aspects of this characterization was the inclusion of the vision processing. To characterize the performance from the control point of view four types of reference inputs were used: step, ramp, parabola and sinusoid. The responses can be used to identify which subsystems can be optimized.