Wigner distribution based motion tracking of human beings using thermal Imaging

Motion tracking has become an important technology in various applications like surveillance, non-cooperative biometrics, virtual reality, etc. We propose a novel approach for motion tracking using only thermal images. We exploited two types of data association (spatial and temporal data association) in the proposed approach in order to reduce the false decisions. In addition to these, Wigner distribution was applied to the difference image for reducing the fluctuation in threshold image in terms of false object detections. The results obtained with this algorithm when applied to six image sequences of about 2500 images each, show its robustness for person tracking in thermal images.