基于非下采样Contourlet变换的红外小目标检测方法

Small Infrared Target Detection Based on Nonsubsampled Contourlet Transform

对存在背景干扰和噪声情况下的红外小目标检测方法进行了分析,提出了一种时空结合的红外小目标检测算法。首先根据背景图像变化较慢的特点,运用相邻帧相减以减少背景和噪声的干扰,接着对残差图像进行非下采样Contourlet变换,利用非下采样Contourle分解后子图像的特性抑制剩余的背景并消除噪声,提高了目标信噪比,最后通过对得到的增强图像进行自适应阈值分割处理,检测出了小目标。仿真实验结果表明,与基于小波变换的红外小目标检测算法相比,该方法能够较精确地检测出红外小目标,具有较高的检测率和较小的平均虚警数。

The small infrared target detection methods used in the presence of background interference and noises are analyzed.A small infrared target detection algorithm combining temporal domain with spatial domain is put forward.First,the interference from background and noises is reduced by using conjoint frame subtraction because of the slow change of background.Then,the nonsubsampled Contourlet transform is implemented for the residual image so as to eliminate the residual background and noises and enhance the signal-to-noise ratio of the small targets.Finally,the enhanced image is processed by using the adaptive threshold and the small target is detected.The simulation result shows that compared with the small infrared target detection algorithm based on wavelet transform,this method can more precisely detect small infrared targets and has a higher detection effectiveness and a lower average false alarm rate.