基于Contourlet和SVR的红外弱小目标检测方法

Detection of Small Infrared Target Based on Contourlet and SVR

本文通过分析红外图像中的小目标和背景特征,提出了基于Contourlet变换和支持向量回归(SVR)自适应滤波器的检测方法。该方法首先采用Contourlet变换抑制背景杂波;然后用基于SVR的自适应滤波器对高频系数进行滤波处理,可以大大提高图像的信噪比;最后根据目标运动的连续性和轨迹的一致性检测小目标,并利用流水线结构的序列处理方法进一步提高检测性能。最终实验结果表明:该方法可显著提高红外小目标的检测概率,实现远距离弱小目标的检测。

By analyzing the characteristics of small infrared target and background, a new detection method based on contourlet and the support vector regression（SVR） is proposed. First, the contourlet is used to suppress the background clutter. Then, the SVR is adopted to process the high frequency coefficients. As a result, the image signal-to-noise ratio （SNR） is greatly improved. Finally, the small target is detected based on the continuity of the target and the consistency of trajectory. And the detection performance is improved by pipelined sequence processing. The experimental results show that the detection method can significantly increase the probability of infrared target detection to achieve long-range dim target detection.