一维均值反差滤波的小目标快速检测算法

Algorithm of small target detection based on one dimensional average gray absolute difference maximum

复杂背景下低信噪比弱小目标的自动检测是当今目标自动探测研究尚未解决的一个难题。将一维均值反差作为一种不相似性度量应用于小目标检测的前期滤波处理中,有效地增强小目标信息、抑制了复杂的背景和噪声,并结合背景预测原理,实现了对小目标的快速检测。仿真实验表明:该滤波算法大幅提高了小目标图像的信噪比,保证了利用背景预测原理检测小目标的准确性;与基于二维均值反差的滤波方法相比,该方法对小目标形状的适应性更强,速度更快。

It is an unfathomed and difficult problem that weak and small targets are automatically detected in complicated background and low SNR（signal noise ratio）.One dimensional average gray absolute difference maximum（1-D AGADM） was presented as the dissimilarity measurement between target and background region.The filter based on 1-D AGADM was used as preprocessing method for small targets detection.It not only enhanced small target,but also suppressed the clutter and noise.Then,the theory of background forecast was utilized to detect small targets.Experiments indicate that the presented filter greatly improved the SNR of small target image,and ensured the veracity of background forecast algorithm in the following detecting small target.Moreover,compared with the 2-D AGADM filter,the novel filter can adapt to the shape of target better and the processing speed is faster.