Sports Video-based Vehicle Detection And Follow-up Study

With the rapid development of our economics, the city's population and the number of vehicle both grow greatly, so do the vehicle flows. So, the problem of vehicle jam becomes serious. Therefore it is required to study the problems of city traffic and find the solution. On the other hand, with the development of computer hardware and computer vision technology, a vision-based traffic monitoring system has become possible. The main purpose of the vision analysis is to detect, recognize and track object, and then try to understand and describe the action of objection. The real-time system based on video for detecting vehicle is the key to traffic monitoring system. The thesis mainly discusses the detection of vehicle based on video. It involves in the following several topics, background update and extraction, moving vehicle segmentation, moving vehicle location and detection, and moving object tracking.The thesis, simply introduces the overview, structure, development of Intelligence Transportation System (ITS), and then summarize several techniques of moving object detection, which are in common use at present, discuss their strongpoint and shortcoming respectively, and simple introduces the system about moving object detection and tracking based on video and mainly do research on the moving object detection and tracking methods according to the designs and its needs. The following researches have been done:1) Background update and extraction. Due to light, it is difficult to detect moving object in video by the invariable background. In this thesis, we proposed the novel and self-adaptive and real-time method of background update and extraction. The experimental results show this method has effective whether there has moving objects or not in the original background.2) The preprocessing of image. In order to remove noise and fill the hole in the region, it is adopted morphologic erosion and dilation methods in the segmented image. Otherwise, it is adopted morphologic shaper method in gray image to enhance the edge information. 3) Moving vehicle detection. It proposed a method from coarse to fine to locate moving objects. Firstly, the region can be coarse located using the horizontal and vertical projection in the segmented image, and then the edge can be detected by multi-scale morphologic shaper operator in gray image after segmentation; finally, the region of moving object can be accurately located by the horizontal and vertical edge projection.4) Tracking moving object. It is proposed a based- center of gravity of region method to track the moving object. The center of gravity of region can be calculated, and the region is signed, and then the moving objects can be tracked by the method of center of gravity of region.Very promising experimental results are provided using real-time video sequences, experimental results have confirmed that the proposed method is efficient and reliable, whether one moving object or multi-moving objects, whether running in straight line or curving road in sequences, and its computer cost can also satisfy the traffic surveillance systems.The cooperation of multi-subject and unremitting effort is the essential condition for automatic moving object detection and tracking and recognition and its practical application. With long term researching and studying, the era of intelligent acquisition and processing and application moving object detection and tracking and recognition will by all means come true!...