IMM-PF visual tracking based on multi-feature fusion

To overcome the tracking problems caused on the one hand by the complex environment and target's appearance changes and on the other hand by maneuvering motion, tracking algorithm was proposed based on multi-feature fusion for particle filter to construct a robust appearance model, and a multi dynamic model under the interactive multiple models (IMM) to describe the target's motion. The proposed method is tested on a variety of videos that present different conditions. Experiences and comparisons conducted show the robustness of our methods in different tracking conditions.