【摘要】 针对具有强机动性的扩展目标跟踪问题，不仅需要估计目标的运动状态，还要准确识别目标机动过程中的形状变化。由于目标机动和模型的复杂所导致的高维非线性问题，提出了一种基于随机超曲面模型的扩展目标方向自适应跟踪算法。通过使用输入估计检测器对目标机动进行判断，再利用目标机动的大小修正形状参数的先验，其次基于先验信息更新量测，并结合无迹卡尔曼滤波算法，实现对机动扩展目标的状态跟踪和形状识别。利用均方根误差和quasi-Jaccard距离分别对目标质心位置的跟踪质量和形状的跟踪性能进行评价。仿真实验结果证明了该算法的有效性。更多还原

【Abstract】 The extended target tracking problem with strong maneuverability requires not only estimation of the target’s motion, but also accurate identification of changes in the target’s shape during maneuvering. Due to the high-dimensional nonlinearities caused by both target maneuvering and model complexities, this paper proposes an extended target maneuvering direction adaptive tracking algorithm based on a random hypersurface model(RHM). This algorithm uses an input estimation(IE) chi-square detector to judge the target’s maneuvering, and corrects the prior shape parameters based on the magnitude and direction of the maneuver, the measurement is updated based on prior information, and the unscented Kalman filtering algorithm is combined to achieve tracking of the state and shape recognition of maneuvering extension targets. The root mean square error(RMSE) and quasi-Jaccard distance are used to evaluate the tracking quality of the target centroid position and the tracking performance of the shape, respectively. Simulation results demonstrate the effectiveness of the proposed algorithm.更多还原