基于图像滤波与Hough变换的红外弱小目标检测  

Infrared Small Target Detection Based on Image Filtering and Hough Transform

作　　者：汪烈兵 姜雄飞 石春光 李慧冲 马晓龙 WANG Liebing;JIANG Xiongfei;SHI Chunguang;LI Huichong;MA Xiaolong(PLA 66069 Troops,Luoyang 471023,China)

机构地区：[1]中国人民解放军66069部队,河南洛阳471023

出　　处：《红外技术》2020年第7期683-687,共5页Infrared Technology

摘　　要：针对红外传感器成像信噪比低且易受噪声、背景杂波干扰的问题,本文提出了一种基于最大中值滤波与Hough变换的红外弱小目标检测方法。该方法首先对序列图像中的每一帧进行最大中值滤波,滤波输出与原始图像相减用以预测目标可能出现的区域。对于目标可能出现的区域使用均值滤波,反之则使用最小值滤波来抑制背景结构。然后将滤波后的序列图像直接叠加得到合成图像,采用有灰度权重的Hough变换实现弱小目标的轨迹检测。实验结果表明该方法可以实现对信噪比接近1的弱小目标的检测。同时,背景的运动对该方法的检测效果没有显著影响。To solve problems caused by infrared image sensors,such a slow signal to noise ratio,background clutter interference and sensor noise,an infrared small target detection method based on max-median filters and Hough transform is proposed.This proposed method predicts a candidate target region using the method of Max-Median filtering with a single image frame.The candidate target region is processed by a mean filter and the residual region is processed by a Min filter.Then,the weighted Hough transform is used to detect targets with the accumulation of multiple image frames.Experimental results demonstrate that this proposed algorithm can detect small targets with a signal to noise ratio of approximately 1.The background movement does not have a significant impact on the effectiveness and robustness of the proposed method.

关 键 词：红外图像 弱小目标检测 最大中值滤波 HOUGH变换 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]