自动检测和跟踪红外图像中的弱小目标在现代预警和探测系统中非常重要。针对现有的检测方法因受到复杂云层和地面自然背景的干扰, 导致系统虚警率较高和探测概率较低的问题, 提出了一种基于多尺度广义模糊算子的红外图像复杂背景抑制方法。首先利用非下采样轮廓波变换方法将图像进行分解, 获取不同尺度和方向的包含小目标和背景杂波的子带图像。然后, 通过应用模糊非线性背景抑制算子将红外图像中小目标和背景杂波系数差值拉大, 同时将相关性较强的杂波系数做平滑处理以达到抑制背景和增强小目标的目的。多组包含真实和模拟图像序列的实验表明所提方法优于其他方法, 特别是对于包含云层和地面复杂背景的红外图像。

Automatic detection and tracking of dim and small targets is very important for early warning and detection system. The existing detection methods usually show high false alarm rate and low detection probability due to the interference of complex clouds and ground background, based on which, a new method of complex background suppression for infrared images was proposed based on multi-scale generalized fuzzy operator. Firstly, the non-subsampled contourlet transform was adopted to decompose the input infrared images to extract sub-band images which contain small targets and background clutter in different scales and directions. Then, the difference of coefficients between small target and background clutter in infrared image was enlarged by using fuzzy nonlinear background suppression operator, and the correlated clutter coefficients were smoothed to suppress the background and enhance the small targets. Several groups of experimental results demonstrate that the proposed method has better effect on background suppression, especially for infrared images containing clouds and complex ground background.