Tracking humans in mutual occlusion based on game theory

Tracking humans in video is a challenge because of data association and mutual occlusion problems. This paper presents a Gaussian mixture probability hypothesis density-based visual tracking system with game-theoretical occlusion handling. First, a two-step occlusion reasoning algorithm is proposed to determine the occlusion region. Then, an improved spatial color appearance with interferences by other targets within the occlusion region is modeled. Finally, an n-person, non-zero-sum, non-cooperative game is constructed to bridge the joint measurements estimation and the game. The individual measurements within the occlusion region are regarded as the players in the constructed game competing for maximum utilities using certain strategies. The Nash equilibrium of the game is the optimal estimations of the locations of the players within the occlusion region. Experiments conducted on publicly available videos demonstrate the good performance of the proposed occlusion handling algorithm. © 2013 IEEE.