A computer vision based camera pedestal’s vertical motion control

Traditional camera pedestals are manually operated. Our long term goal is to construct a fully autonomous pedestal system which can respond to changes in a scene and mimicking the human camera operator. In this paper, we discuss our experiments to control the vertical motion of a pedestal by leveling its position with a human head or a tracked hand-held object. We describe a set of computer vision methods used in these experiments, including the head position tracking using Gaussian mixture model (GMM) of the foreground blob and hand-held object tracking using continuously adaptive mean shift (CAM-shift) with motion initialization. We also discuss the application of Kalman filter and showing its effect in the reduction of the number of jittering pedestal motions.