Robust People Counting in Complicated Situations

Robust People Counting in Complicated Situations

  People counting is a basic yet important aspect in intelligent video surveillance. In some places full of moving pedestrians or objects, it is a real challenge for traditional image processing approaches. This paper addresses the issue of how to enhance the robustness of people counting in two complicated situations, one is to count individuals mixed with objects and the other is to count people in a crowded scene. Based on modeling method and machine learning, we employ a new way to tackle the problem, which reduces the error rate and the miss rate dramatically. Experimental results have demonstrated the robust function of the system.