OF-SMED: An optimal foreground detection method in surveillance system for traffic monitoring

Foreground detection is a key procedure in video analysis such as object detection and tracking. Several foreground detection techniques and edge detectors have been developed until now but the problem is, usually it is difficult to obtain an optimal foreground due to weather, light, shadow and clutter interference. Background subtract is a common method in foreground detection. In background subtract noise appears at fixed place, when it is used to deal with long image sequence there may be much accumulate error in the foreground. In OF (Optical Flow) noise appears randomly and this covers long distance over long period of time. Optical flow cannot get rid of the light influences which result in background noises. To overcome this SMED (Separable Morphological Edge Detector) is used. SMED has robustness to light changing and even slight movement in the video sequence. This paper proposes a new foreground detection approach called OF and SMED which is more accurate in foreground detection and elimination of noises is very high. This approach is useful for efficient crowd and traffic monitoring, user friendly, highly automatic intelligent, computationally efficient system.