A crowd-filter for detection of abandoned objects in crowded area

Terrorist attacks have become a critical threat of public safety; especially, explosive attacks with abandoned packages are repeatedly concentrated on public places such as MRT stations. Hence, establishing a surveillance system with high-tech appliances to against terrorism is a critical issue nowadays. Unlike most of the existing object detection schemes which mainly deal with static or monotonic surveillance scenarios, abandoned object detection in crowded area becomes much important on against anonymous explosive attacks. In this paper, a crowd-filter for abandoned object detection in crowded area is proposed. According to the proposed vertical scan line filter, foreground moving pedestrians can be isolated from a crowded scenario; therefore, abandoned objects can be detected and alarmed timely. Experimental results demonstrate that the proposed crowd-filter can successfully restores a crowded surveillance video to a clean scene and then identify an abandoned object using simple background subtraction technique.