Multi-person vision tracking approach based on human body localization features

This paper presents a multi-person vision tracking approach based on human body localization features to address the problem of interactive object and in home monitoring scenario. Firstly, model is used obtain 3D position body, which then construct motion Kalman filter method. At same time, appearance constructed by fusing color histogram oriented gradient better characterize body. Secondly, observation measure similarities between state sequence historical frame result current frame, cost matrix obtained. Thirdly, Hungarian maximum matching algorithm employed match each frames, exception detection mechanism simultaneously further reduce probability failure. Finally, verification platform was constructed, achieved average accuracy 96.6% case overlapping, occlusion, disappearance, appearance; this verifies feasibility effectiveness proposed