基于帧间特征点匹配的红外弱小目标检测

高空下、视复杂背景下弱小目标的检测一直是红外弱小目标跟踪的难点,提出了一种基于帧间特征点匹配的红外弱小目标检测的方法,将复杂背景下的动态弱小目标检测问题看作是帧间复杂背景的运动估计补偿问题,消除了背景杂波对红外目标的影响,进而达到了抑制背景的目的。做出了该算法与现在常用的频域高通滤波、形态学Top-hat滤波两种小目标检测算法的对比,并将目标局部信杂比和目标检测的虚警率作为算法的有效性评价指标。试验结果表明,通过准确的地补偿复杂背景的帧间位移量,再结合帧间差分的方式,后红外弱小目标检测的信杂比提升了2倍以上,检测得到的虚警率低于20%。

Small Target Detection in Clutter Used Feature Point Matching between Frames

The detection of small targets in complex background at high altitude has always been the difficulty of tracking infrared weak targets. A small target detection adopting feature point matching between frames is proposed in this paper. The problem of small target detection in complex background is regarded as the problem of complex background motion estimation compensation of different frames, which can eliminate the effect of background clutter on infrared target and achieve the purpose of background suppression. Then the method is compared with several commonly-used small target detection algorithms like high pass filter and Top-hat filter. The target local signal to noise ratio and false alarm rate are taken as the evaluation index. The experiment results show that after accurate background compensation and frame difference, the target signal-to-noise ratio can increase by more than 2 times and while the detection false alarm rate is lower than 20%.