时空域滤波方法在红外弱小目标检测系统中的应用

Application of temporal-spatial domain filter in infrared small and dim targets detection system

针对远距离红外弱小目标低信噪比的特点,提出一种空域滤波与粒子滤波相结合的目标检测方法。首先在空域进行形态滤波Tophat变换,抑制背景和背景杂波,得到了滤除背景的目标图像,再通过滑窗累积方法进一步增强目标,提高目标的信噪比,增大检测概率。然后根据目标时域运动的连续性和规则性,采用基于粒子滤波预测的邻域目标检测判决方法,达到弱小目标检测的目的。实验结果表明:该方法能够有效地检测出幅度信噪比约为1.7的弱小目标,对背景变化情况下的弱小目标检测具有较好的适应性。

A novel dete ction algorithm combined spatial domain filter with particle filter was proposed for detecting small and dim targets with low SNR in infrared images in long distance . At first, the morphologic filter Tophat transform was achieved in the spatial domain to compress background and background clutter, the target image without background clutter was obtained. Then, the method of slipe-window accumulation was adopted to enhance the target further, and the SNR of the target was improved, therefore the detection probability could be increased. Finally, according to the continuity and regulation of the targets movement in temporal domain, an algorithm based on neighborhood judgment of particle filter forecasting was put forward to achieve the goal of small and dim targets detection. The experimental results show that the algorithm can detect small and dim targets effectively with SNR of 1.7, and has well adaptability for small and dim targets detection under variational background.