A Multiple Motion Model Tracker Handling Occlusion and Rapid Motion Variation

We propose a new tracking method capable of handling occlusions and non-constant target motion. This is achieved using multiple simple motion models, learned at different temporal scales and combined to capture possibly complex motion patterns. These motion models are learned online in a computationally inexpensive manner. Reliable recovery of tracking after occlusions is achieved by extending the bootstrap particle filter to propagate particles at multiple temporal scales, guided by the simple models. In complex environments targets can display changes in direction or speed unaccounted for by standard polynomial motion models. To demonstrate the generality of our framework and accommodate these changes, the proposed method is also applied to a more flexible, two-stage motion model. Extensive experiments have been carried out on both publicly available benchmarks and new video sequences. Results reveal that the proposed method successfully handles occlusions and a variety of rapid changes in target motion.