Particle Filters for Target Tracking Using Vision Data for Micro-Air Vehicles
Myungsoo Jun, David E. Jeffcoat, Johnny H. Evers · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2006
This paper considers a target tracking problem using vision data for micro-air vehicles (MAV’s). Most previous visual tracking algorithms are not appropriate for rapidly ∞ying MAV’s with limited computational abilities. The proposed target tracker consists of two parts | an image processor and a particle fllter. The image process detects a target with high contrast against background and yields bearing measurements of the target. Since the segmentation image process does not always successfully recognize the target due to noise in the images, the particle fllter compensates and estimates the location of targets in the presence of noise and uncertainty. Experiments show that the proposed visual tracking algorithm performs well in the presence of noise and runs in real time adequate for a rapidly ∞ying MAV.