基于OWRR-PF的机载图像中导弹跟踪方法

针对利用图像信息实现飞行器的导弹预警跟踪中导弹运动2D建模、运动模型非线性和目标干扰非高斯等问题,研究了比例导引下导弹在三维空间中的运动在成像面上的运动模式,建立了比例制导导弹的2D投影运动状态模型.由于模型中速度和加速度等主要物理量的非线性,并考虑导弹运动过程中受到的风向、风力、气旋和气流等随机干扰的非高斯性,采用粒子滤波方法实现导弹跟踪;并针对粒子滤波在跟踪过程中存在的粒子退化问题,提出有序权值残差重采样粒子滤波(OWRR-PF,Ordered Weight Residual Resampling Particle Filter)方法,该方法缓解了粒子退化问题,提高了跟踪的准确度.利用所建立的导弹运动模型进行连续视频试验,与标准粒子滤波相比跟踪精度提高了70%左右,与残差重采样粒子滤波方法相比跟踪精度提高了15%左右.

Method of missile tracking in images based on ordered weight residual resampling particle filter

There were many problems existing in the missile alarm and tracking for aircrafts,such as 2D missile motion modeling,the nonlinearity of the motion model and the non-Gaussian of the interference.The motion pattern of the proportional navigated missile from the 3D space to the camera imaging plane was studied and the motion model of the 2D projection of the proportional navigated missile was established.Since the velocity and acceleration in the model was non-linear,and the random interference(e.g.,wind direction,wind power,cyclone,air flow) encountered by missiles was also non-linear,the particle filter was used to track missiles.As for the particle degeneration problem in the tracking procedure,the ordered weight residual resampling particle filter(OWRR-PF) method was presented to release the particle degeneration problem and improve the tracking accuracy.Experimental results demonstrate that the tracking accuracy by using OWRR-PF method was improved by 70% than the standard particle filter,and 15% than the residual resampling particle filter.