基于自适应Gabor滤波的红外弱小目标检测  

Dim Infrared Target Detection Based on Adaptive Gabor Filter

作　　者：史漫丽[1] 凌龙[1] 吴南[1] 原娜 SHI Manli;LING Long;WU Nan;YUAN Na(Beijing Institute of Space Mechanics&Electricity,Beijing 100094,China)

机构地区：[1]北京空间机电研究所,北京100094

出　　处：《红外技术》2018年第7期632-637,共6页Infrared Technology

基　　金：国家自然科学基金(61775030;61571096);中国科学院光束控制重点实验室基金(2017LBC003)

摘　　要：针对复杂背景下的红外弱小目标检测,本文提出了一种改进的Gabor滤波的红外弱小目标检测方法。该方法在背景预测算法的基础上,通过构造Gabor核函数来自适应确定背景预测系数。该方法利用了更多的图像局部特性信息,使用对比度尺度模型和强度尺度传播模型分别确定Gabor核函数的两个轴,解决了Gabor滤波算子不能自适应调整滤波系数的问题。通过与传统的小目标检测方法的比较实验结果表明,本文方法能有效保留图像的边缘信息,能有效地突出目标,抑制背景杂波,提高了对红外弱小目标的检测能力,效果明显优于传统方法。To realize infrared dim target detection in a complex background,an improved infrared dim target detection method based on the Gabor filter is proposed in this paper.Here,the Gabor kernel function is used to determine the coefficient of the background prediction algorithm;moreover,the problem of the filter operator not adjusting to the filter coefficient adaptively is solved by the Gabor filter operator.To use more local characteristic information of the image,the contrast scale model and intensity scale propagation model are introduced,and the two axes of the Gabor kernel function are determined.A comparison of the proposed method with the traditional small target detection methods demonstrates that the proposed method can effectively retain the edge information of the image,suppress the background clutter and highlight the target,and improve the single frame detection ability of dim targets.The effect achieved by our proposed method is considerably better than the traditional method.

关 键 词：GABOR 背景预测 红外弱小目标 目标检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]