Moving Object Detection and Tracking for Video Surveillance System

Abstract

Video surveillance has long been in use to monitor security sensitive areas such as banks, department stores, highways, crowded public places, University campus and borders. The video outputs are monitored by human operators from a control room and are usually saved to tapes for later use only after a forensic event. The smart surveillance system makes this job easier. The making of video surveillance systems smart requires fast, reliable and robust algorithms for moving object detection and tracking. In this thesis, a smart visual surveillance system with real-time moving object detection and tracking capabilities is presented. The system operates on both color and gray scale video imagery from a stationary camera. It does handle detect and track the path of object in indoor as well as outdoor environments and under changing illumination conditions. This thesis is divided in two modules: Object Detection and Tracking. To achieve moving object detection goals, three processing levels are proposed: segments moving objects from the background, reduce noise and video analysis to extract meaningful objects and their features (area, center of mass etc.). Adaptive Background Subtraction method is used for segments moving objects from the background image, which is capable of adapting to dynamic scene conditions. Morphology operation (Erosion and dilation) are applied to remove noise. In last level, The filtered foreground pixels are grouped into connected regions and are labeled we extract the meaningful object and their features. After obtaining binary image, object association between consecutive frames is achieved by a tracking method. To achieve object tracking goals, correspondence base matching algorithm is used. This system does not track object parts, such as limbs of a human. It does detect and track the path of moving objects as a whole from video scenes.