基于Kalman滤波的红外弱小目标检测前跟踪算法

红外弱小目标检测跟踪问题具有重要的军事意义和广阔的应用前景,检测前跟踪算法是解决这一问题的有效途径。提出了一种基于Kalman滤波的检测前跟踪算法：首先对红外图像进行形态学top-hat算子滤波预处理;然后利用恒虚警率阈值提取单帧候选目标,并利用目标灰度模板进行灰度核密度估计,初步剔除大部分虚假目标,累积处理若干帧后,利用Kalman滤波器筛选出最优轨迹;最后依据一定的判断准则从当前帧候选目标中提取出真实目标。与一种典型的基于管道滤波的算法进行对比,仿真实验结果表明,该算法对目标运动速度和信噪比的变化有较强的适应能力,同时能用于目标遮挡或消失等情况。

Track-before-detect Algorithm for Infrared Dim Target Based on Kalman Filter

The problem of IR dim target detection and tracking is important for military technology,and applies in many regions expectly.Track-before-detect(TBD) method is the best way for solving this problem.A track-before-detect algorithm is proposed based on Kalman filtering.First,IR image is preprocessed by morphology top-hat operator.Second,thresholded in single frame according to constant false alarm rate(CFAR),computes gray level kernel density estimation to target template,removes most false targets preliminarily,and accumulates several frames.Finally,filters to likelihood trajectory used Kalman filtering,and estimates real target by optimization rule from current frame.A representative track-before-detect algorithm based on pipeline filter is contrasted.The simulation results show that the proposed algorithm is efficient to change the velocity and signal-noise-ratio(SNR),and can apply in the case of disappeared.