Joint Tracking of Moving Objects With EO and IR Cameras

摘要

Tracking moving objects, especially human objects in surveillance systems has attracted considerable research attention. This study proposes a novel joint Electro-Optical (EO) and Infrared (IR) cameras tracking approach by employing particle filter. A centroid-based detection technique is used to discover potentially moving objects and obtain the coordinate data. Once moving targets are detected, both EO and IR features are combined to extract object templates for sampling particles. Statistic information of a blob centered at current particle and likelihood of each pixels in terms of foreground, background and occlusion components are obtained, to determine and update importance of each particle and handle temporary occlusion. Hence, particles which can provide accurate prediction are assigned with higher weights. Simulations have been conducted to validate the proposed method.