Recognition process using feature data fusion for imaging system

We present a system that recognizes moving targets based on feature data fusion. This system consists of a two dimensional CFAR process to reduce background clutter, a motion vector process to detect moving objects, a gate process and a target detection process. It can detect and track moving objects efficiently under complex backgrounds. It can be applied to automatic surveillance systems for outdoor security, traffic control, crime prevention and so on. We use segment information such as positions of edges obtained by the differential process and gravity of area obtained by the binary process. The above information is evaluated as features of segments. By applying the proposed system to outdoor surveillance, its effectiveness is evaluated.