CAJ | 학술논문

为实现对海上运动目标的实时跟踪,克服跟踪效果易受到严重遮挡影响的缺点,建立了一套实时跟踪系统,并且结合目标的运动信息与新的模型更新策略,对均值漂移与卡尔曼滤波器相结合的跟踪算法做出了改进.当运动船只被遮挡的比例较大时,先用估计出的目标速度矢量更新卡尔曼滤波器,并用相应的模型更新策略更新目标模型以提高模型相似性度量的精确性,再单独利用滤波器进行跟踪,取得了较好的跟踪效果.实验结果表明,该系统可以较好地实现海上运动目标的跟踪,并且改进后的算法具有良好的实时性和鲁棒性.
위실현대해상운동목표적실시근종,극복근종효과역수도엄중차당영향적결점,건립료일투실시근종계통,병차결합목표적운동신식여신적모형경신책략,대균치표이여잡이만려파기상결합적근종산법주출료개진.당운동선지피차당적비례교대시,선용고계출적목표속도시량경신잡이만려파기,병용상응적모형경신책략경신목표모형이제고모형상사성도량적정학성,재단독이용려파기진행근종,취득료교호적근종효과.실험결과표명,해계통가이교호지실현해상운동목표적근종,병차개진후적산법구유량호적실시성화로봉성.
To accomplish the real-time tracking for the marine moving object and overcome the defect of the susceptibility to severe oc-clusion, a set of real-time tracking system for the moving object at sea is designed. In virtue of the motion information and a new kind of model update strategy, some improvements are made for tracking algorithm combining the mean-shift with Kalman filter. When the moving vessel is largely blocked, Kalman filter is updated by the prior estimated velocity vector, and the accuracy of model similarity measurement is improved by utilizing the model update strategy, then the filter is exploited to implement tracking. The algorithm achieved good tracking effect. Experimental results show that the system implement marine moving object tracking with good real-time perfor-mance and robustness.