Moving Object Detection And Tracking Algorithm For Video Sequence

Dynamic visual information takes the most part of environment information that is perceived by human beings. Therefore, perceiving the dynamic visual information in the environment has become an important research field in computer vision. Moving object detection and tracking is one of the most important issues in applied vision and moving image coding and has wide applications in many fields. Its applications include fields like robot navigation, vision-based supervision, security surveillance, medical image analysis, industrial automation, video processing, etc.The study object of this paper lays an emphasis on the research of moving objects detection and tracking in image sequence obtained by a stationary camera. The basic concepts of OpenCV and digital image processing are introduced firstly. The basic concepts include color model, image pretreatment and morphology. Moreover, the effect of impact processing is given. Secondly, for moving object detection, three methods, including differential between adjacent frames, background subtractction and optical flow, are discussed. Through comparison and analysis, the merits, weaknesses and application field of each method are concluded. Especially, while introducing the background subtraction, the paper also compares the differences of the signal Gaussians model and mixture Gaussians models. At the same time, this paper presents a new background modeling method based on codebooks. Finally, for moving object tracking, three methods, including Kalman filter, Mean Shift and CamShift, are introduced concretely. Based on these, a new tracking method is presented..An Kalman filter is used to predict the moving object’s position in realtime at the next moment, so that CamShift can search and object matching in a less range, thus reducing calculating in a certain extent. Using Visual C++and OpenCV, the method can detect and track moving object accurately.