基于塔型对偶树方向滤波器组的弱小目标背景抑制方法

复杂背景抑制是红外监视告警系统探测远距离目标的一个难题，提出了一种将塔型对偶树方向滤波器组与改进的视网膜皮层理论相结合的弱小目标背景抑制方法。首先，采用塔型对偶树方向滤波器组对图像进行多尺度、多方向分解，提取图像的多尺度和方向细节特征；然后，根据目标和背景杂波信号的差异，通过应用改进的视网膜皮层理论公式调整分解后的各子带系数，从而将红外图像中弱小目标信号和背景杂波分离，达到抑制背景的目的。与二维最小均方误差和最大中值滤波方法相比较，实验结果显示该方法能有效地检测出信杂比在1.6以上的目标。

Background Suppression for Dim and Small Targets Based on Pyramidal Dual-tree Directional Filter Bank

The complex background suppression is difficult for infrared warning system detecting long- distance target. A background suppression method based on pyramidal dual-tree directional filter bank was proposed for some dim and small targets and retina theory was introduced to improve its ability. Firstly, the filter bank was adopted to decompose the input infrared image into multiple scales and directions and extract the image's features. Then, according to the difference between target and background clutter signal, the decomposed sub-band's coefficients were adjusted by using improved retina theory equations to suppress the background details and enhance target information. Several groups of experiment results demonstrate that the proposed method can detect the target with SCR>1.6 effectively, compared with several classical background suppression method, such as two-dimensional least means square (TDLMS) and max median (MMed) filters.