【摘要】 提出一种基于雷达/红外传感器神经网络融合的机动目标跟踪算法,利用神经网络的非线性逼近能力,将神经网络与卡尔曼滤波器相结合构成一个非线性估计器,该算法可以对来自红外成像传感器的补充信息加以充分利用,进行机动检测,把计算负荷转移到神经网络,在改善跟踪性能的同时又保持跟踪滤波的计算结构尽可能简单。仿真结果表明所提出的跟踪滤波算法在跟踪应用上优于一般的非线性估计算法,它最明显的优点就是减少了数字计算上的复杂性,提高了跟踪算法的快速性。更多还原

【Abstract】 A maneuvering target tracking algorithm based on Radar / Infrared sensor neural network fusion is presented. A neural network with a Kalman filter is characterized with a nonlinear tracking filter, which enables to make fully use of the image-based additional information for maneuvering detection and keeps the simplicity of the algorithm for the part of its computation load is transferred to the neural networks. Simulation results show that the proposed algorithm has significant advantages over the common nonlinear estimation algorithms in tracking applications for its reduction of computation complexities and its improvement of calculation speed.更多还原