MANEUVER DETECTION AND OBJECT TRACK PREDICTION USING TELEVISION SURVEILLANCE SYSTEM
Central Scientific Research Institute ”Kometa”, Scientific Design Center of Optoelectronic Surveillance Systems; Director
L. D. Vilesov
St. Petersburg State University of Aerospace Instrumentation, Department of Radioelectronics and Television;
V. I. Mojeiko
Central Scientific Research Institute ”Kometa”, Scientific Design Center of Optoelectronic Surveillance Systems; Head of Department
T. Y. Fissenko
Central Scientific Research Institute ”Kometa”, Scientific Design Center of Optoelectronic Surveillance Systems; Senior Scientist
Read the full article
Abstract. Methods of object coordinates measurement by television surveillance system are considered. The observation data errors are filtered out by means of an adaptive Kalman filter, adaptation is related to possible change of object trajectory at an unknown instant and subsequent loss and appearance of the object (e.g., because of a cloudy cover). Maneuver of object is detected from velocity estimation after Kalman filtering which is adapted with the use of variation of filter coefficients, while object disappearance switches the filter to trajectory prediction mode. Algorithms of maneuver detection accounting for loss of the object are designed; dynamic and fluctuation errors of trajectory prediction when the object appears from area of occlusion are estimated.
Keywords:
occlusion tracking, adaptive Kalman filtering.
