Moving object tracking using camera motion corrected Kalman filtering

Displays the improved performance of an α-β-γ tracker which is obtained by compensating for the camera motion. The authors consider the effect of data drop out on the filter. They review the α-β-γ trackers, and define the transformation between the global and camera coordinate frames, which, together with the point-centered projection onto the image plane, define the corrections which compensate for changes of yaw, pitch and roll. Finally the authors present the results of the compensated filter using synthetic data