Traffic statistics based on moving target tracking technology
|School||Kunming University of Science and Technology|
|Course||Communication and Information System|
|Keywords||ITS Detection and tracking Hausdorff distance Kalman filter Traffic flow|
With the developing of the cities, traffic becomes a very serious problem in modern society. In order to solve the problem, the computer science and communication technologies were applied into vehicle traffic controlling and monitoring to build the Intelligent Traffic System (shorten as ITS). Tracking moving targets is one of the major ITS technologies, and obtaining real-time traffic information is very important for ITS. Based on object tracking, we could obtain many useful information, for example:traffic flow which is a very important data of the ITS and has a vast perspective of application.In this paper, in order to extract accurate traffic information for research purposes, to the image sequence moving target detection and tracking moving objects and traffic statistics, such as technology at the core, an in-depth study is made.About the research of moving object detection, firstly, several main algorithms of detection are researched, comparing the benefits and weaknesses. Then, For background-difference method, about how to obtain and refresh the background, a background extracting method based on gray-level classified is designed in this paper. Finally, after preliminary objects detection, the paper propose a shadow detection method based on measure the flatness of the divided image.About the research of moving object tracking, firstly, the paper discuss several main algorithms of tracking and then the paper research matching algorithm based on Hausdorff distance in detail. In order to reduce quantity of tracking and achieve the purpose of multi-target tracking and traffic flow counting, the paper design a object tracking algorithms combined Hausdorff distance matching with Kalman filter to forecast the object moving location.For the strategy of moving objects detection and tracking, the paper have sequence images experimented and the result shows that this method performs well.