Fusion of two different motion cues for intelligent video surveillance

Abstract

Detecting the presence of people and suspicious objects are the essential tasks for security surveillance. This paper presents a new real-time system for this purpose. Two motion cues from background subtraction and temporal differencing are employed to not only get reliable motion detection but also identify detected objects in the scene. A fuzzy reasoning technique is developed to detect and locate motion objects from vertical projection of motion cues. The background model is updated on both pixel level and region level to adapt both slow illumination changes and sudden extraneous events. This new background maintenance technique makes the system able to work under varying environments. The system has been run round the clock in real scenes and the performance is very promising.