A gradient-based foreground detection technique for object tracking in a traffic monitoring system

Moving object detection is an important process in video surveillance and monitoring system. A conventional approach to moving object detection in video surveillance application is background subtraction. Information about background can be determined by various techniques. However, many existing background modelling techniques suffer in uncontrolled environments. This paper presents a new background modelling approach based on edge information. The proposed approach is robust to changes in outdoor environments where light can change randomly. Furthermore, edges require significantly less computing resource and result in faster processing time relative to intensity-based approaches. Experimental results show that, compared to intensity-based approaches in outdoor environments, high detection accuracy has been achieved with significantly less computing power.