A Robust Method for New Object Detection in Video Surveillance Systems


Abstract Video surveillance systems is a powerful tool for public safety and with the increasing need for more security in airports, banks, schools and other critical environments, the demand for video system is growing rapidly. Sides from the intrinsic usefulness of begin able to segment video streams into moving and background components, detecting moving blobs provide a focus of attention for recognition, classification and activity analysis, making these later processes more efficient since only “moving” pixels need to be considered. In this paper an efficient moving object detection method using modified Horprasert model for video surveillance system is present. The modified dynamic thresholds are able to detect a new object with it's shadow through different video stream in terms of light conditions. It consists of background model, distortion of brightness, color calculations and classification. Four regions are segmented depending on the thresholds foreground (moving object), background, highlight background and shadow. The proposed automatic threshold depends on background computations of brightness and is thus expected to achieved better classification performance.Keywords: Horprasert model; background subtraction; background model; Video Surveillance.