Title :
Detecting Gathering Incident of Video Surveillance Based on Plane Geometry
Author :
Cai, Zhaoquan ; Huang, Han ; Wu, Wenzhong ; Ma, Xianheng ; Hu, Xiuhong
Author_Institution :
Network Center, Huizhou Univ., Guangzhou, China
Abstract :
Gathering is an important incident of video surveillance, possibly reflecting the properties of mass disturbances. However, the incident is more difficult to detect than single-object incidents since it always has multiple detected objects. This paper proposes a plane geometry model to describe the position, trajectory and outline of the detected objects. Gathering incident is checked by a circle in which there are multiple objects. A brief detection algorithm is designed to draw a circle and check whether gathering happens. Finally, several public surveillance videos are tested, and the results indicate the proposed model is efficient, high-detection-rate and robust.
Keywords :
Detection algorithms; Geometry; Humans; Image sequences; Layout; Object detection; Solid modeling; Testing; Tracking; Video surveillance; Computer vision; automatic detection; gathering incident; human behavior understanding; video surveillance;
Conference_Titel :
Machine Vision and Human-Machine Interface (MVHI), 2010 International Conference on
Conference_Location :
Kaifeng, China
Print_ISBN :
978-1-4244-6595-8
Electronic_ISBN :
978-1-4244-6596-5
DOI :
10.1109/MVHI.2010.170