Aircraft tracking using weak detectors with hidden Markov chain

Tracking aircrafts in takeoff and landing procedures is a challenge task, since the appearance of the tacking object is uncertain due to different aircraft types and paintings, and its appearance is varying during the fast moving procedure. In this paper, an approach that uses weak cascade detectors with hidden Markov chain is proposed as a robust real-time aircraft tracker. While cascade detectors have the benefit of fast computation speed, hidden Markov chain model increases the overall performance in reliability and accuracy. Although the performance of the proposed approach can be improved in several ways, the motivation of this approach is to meet the required performance using as low computation cost as possible. It was demonstrated in real-case scenario that it was able to stay above 12 FPS detection frame rate in a 1080P video stream using a single 2.4GHz CPU thread, and successfully tracked all the aircrafts in field test.