Fusion Framework for Maneuvering Vehicle Tracking Based on 3D Model
This paper focuses on the problem of radar and image sensor fusion in vehicles tracking field. For the mobile case, a dynamic model of the vehicle motion is adopted to describe the vehicle motion. Preliminary position estimation of the target is performed based on information from radar sensor, and the more accurate one is re-calculated in a vision window that is established by integrating the preliminary estimation with the image intensity data. A weighted Hausdorff distance is introduced to define the functional relationship between image and three-dimensional (3D) model projection, and the targets orientation measure is achieved by the function optimization. Data fusion algorithm based on square root unscented Kalman filter (SR-UKF), which is more robust to non-linearity than extended Kalman filter (EKF) based one, is proposed to estimate the state parameters of the tracking system. The experiments show that the proposed method can significantly improve the tracking accuracy of the moving vehicle. keywords:information fusion, three-dimensional model, maneuvering vehicle tracking, UKF
Ying Chen Dinghui Wu Zhicheng Ji
School of Control and Communication Southern Yangtze University Wuxi, P.R.China 214122 School of Control and Communication Southern Yangtze University Wuxi P.R.China 214122
国际会议
南宁
英文
2007-07-20（万方平台首次上网日期，不代表论文的发表时间）