全局-局部联合滤波的红外小目标背景抑制方法  

Global filter combined with local filter for infrared small target background suppression

作　　者：莫金花[1] 郗润平[1] 张艳宁[1] 张秀伟[1] 仝小敏[1] 

机构地区：[1]西北工业大学计算机学院陕西省语音与图像信息处理重点实验室,西安710129

出　　处：《中国体视学与图像分析》2011年第3期223-231,共9页Chinese Journal of Stereology and Image Analysis

基　　金：国家自然科学基金(60903126;60872145);中国博士后特别基金(201003685);中国博士后基金(20090451397);西北工业大学基础研究基金(JC201120;JC201148)

摘　　要：背景抑制可有效降低杂波干扰、突显观测目标,是红外小目标检测关键技术之一。通过对现有背景抑制算法的对比分析发现:全局滤波和局部滤波具有一定的互补特性,全局滤波可有效抑制大面积缓变背景,但对起伏背景的边缘区域抑制有限,局部滤波在抑制背景边缘和"脉冲"噪声方面效果较好。基于此,构造了一种全局-局部联合滤波器,结合奇异值分解良好的数值稳健性和自适应性、Robinson Guard边缘滤波器具有保留目标边缘及其内部信息的特点进行背景抑制。通过多组实验分析,并与单一滤波和基于小波与形态学联合滤波的方法对比,结果表明,该方法在信噪比增益和背景抑制程度上更具优越性。Background suppression,which can reduce clutter interference and highlight target effectively,is a key technique in infrared small target detection.It is found that global filter and local filter are complementary.Global filter can work well in slowly changed background region,but it is limited as to the edge region of fluctuated background which can be handled better by local filter.Based on this observation,a global filter local combined with local filter is proposed for infrared small target background suppression.This method takes the advantages of the SVD（singular value decomposition） filter which has good numerical stability and adaptivity,and Robinson Guard filter where the edge and interior information of target can be reserved.Comparison with methods based on a single filter as well as wavelet and morphological combined filter is conducted and the results experiments demonstrate that the proposed method has better performance on SNR gain and background suppression.

关 键 词：背景抑制 奇异值分解 ROBINSON Guard滤波 红外小目标 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]