With the rapid development of economy, the acceleration of urbanization and the generalization of automobiles, many problems have been becoming more serious, such as traffic congestion, frequent accidents, environment deterioration and energy shortage. So intelligent transportation system (ITS) have come into being though utilizing information technique in order to solute the above problems. Image processing and analysis have played an important role inintelligent transportation system, possessing extensive application value. Especially camera have been widely applied to traffic management system(TMS). For a video-based TMS(vTMS), the basic function is automatically to extract real-time traffic parameters, the accuracy of which is depended on the calibrated camera parameters and the used image processing algorithms. So if only camera parameters can be calibrated properly, video-based trafficparameters can be estimated accurately. Therefore, this thesis has mainly researched dynamic camera calibration technology oriented on video traffic flow parameters, which can satisfy the requirement of the exactness of the detectedtraffic parameters by adjusting camera parameters. With this research object, the contributions of the thesis include the following aspects:1.This thesis analyzes research current status and development of camera calibration, and points out existent problems and solutions of camera calibration in the fields of current vTMS.2.This thesis establishes a mathematical model of camera calibration based on traffic video scenes. It mainly makes use of the characteristic of ubiquitous parallel lane markings and the character of parallel lines in projective geometry, in addition deduces the solution approach of camera parameters according to this model.3.This thesis presents a set of lane detection algorithms by integrating a few of existing algorithms in traffic video scenes, subsequently synthesizes and provides a series of image processing and analysis algorithms applied to dynamic camera calibration, synthetically considering various factors which affect the exactness of lane etection.4.This thesis researches a few methods of acquiring traffic parameters about length or location in video scenes. It mainly encircles the calibrated camera parameters to obtain traffic parameters, and the interrelated emphasis is length measurement.5.This thesis presents a dynamic camera calibration system oriented on traffic video parameters. It mainly analyzes and designs the synthetical solution of which from four aspects, such as lane detection algorithm, camera parameters calibration, systemic automation and dynamic.6.This thesis makes some experiments in four aspects, such as mathematical model of camera calibration, lane detection algorithm and acquiring traffic parameter method.
修改评论