DETECTION OF ABANDONED/REMOVED OBJECTS IN VIDEO SURVEILLANCE SYSTEM

This paper proposes a novel method of abandoned object detection and removed object detection. The statistical model is used to establish short-term background and long-term background. Background differencing and coterminous frame differencing is combined to obtain the moving object, and only the non-moving region of short-term background is updated in real-time. Differencing the short-term background and long-term background each 10 frames to obtain the object region, and the edge determinant method and the Bhattacharya coefficient determinant method is combined to determine whether the object is abandoned or removed. The experiments testify that proposed method detects abandoned and removed objects accurately.