Evaluation of Automatic Detection of Humans and Vehicles

This paper describes recent efforts to evaluate and characterize an automated wide-area surveillance system called VISDTA (Video Imaging System for Detection, Tracking, and Assessment). This system has been developed for ground-based detection of humans and vehicles and uses an infrared imager and digital video motion detection (VMD) processing. A description of VISDTA, a discussion of the problems with detecting humans and vehicles, and actual performance data from extensive field tests are provided to characterize the automatic target detection performance. Results from analyses of the data as it relates to environmental factors are also presented, along with recommendations for future efforts in this area.