基于时空高阶累积量与各向异性背景抑制的红外弱小目标检测算法  

Infrared dim target detection algorithm based on temporal-spatial higher-ordercumulation and anisotropic background suppression

作　　者：潘良 邱建林[2] Pan Liang;Qiu Jianlin(School of Computer and Information Engineering,Nantong University of Technology,Nantong 226002,China;School of Computer Science and Technology,Nantong University,Nantong 226002,China)

机构地区：[1]南通理工学院计算机与信息工程学院,南通226002 [2]南通大学计算机科学与技术学院,南通226002

出　　处：《电子测量与仪器学报》2018年第11期118-124,共7页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(61202006);计算机软件新技术国家重点实验室开放课题基金(KFKT2012B29);江苏省科技创新基金(BC2013167)资助项目

摘　　要：为了改善弱小目标的检测准确度,设计了时空高阶累积量耦合各向异性背景抑制的红外弱小目标检测算法。首先,根据红外图像的边缘停止函数,考虑弱小目标与背景的差异,构建各向异性背景抑制方案,以弱化背景干扰;随后,基于传统的高阶累积量,考虑空域特性,利用弱小目标的运动能量来建立一种时空高阶累积量方案,对背景抑制结果中的候选目标进行增强处理;最后,引入尺度空间理论,计算弱小目标的直径,以此改进了管道滤波方案,以充分消除伪目标,从而准确检测弱小目标。测试数据表明,相对于已有的红外目标检测方案,在噪声与杂波环境下,所提方案可以准确地识别出真实目标,具有更高的信杂比增益与背景抑制因子,呈现出更好的ROC曲线。In order to improve the detection precision of the weak and small target in the complex infrared background, this paper designs a weak infrared target detection algorithm based on temporal-spatial higher-order cumulation and anisotropic background suppression. Firstly, according to the edge stopping function of infrared image, the anisotropic background suppression scheme was constructed by considering the difference between dim target and background to weaken the background interference. Then, a temporal-spatial high order cumulate scheme was established by using the motion energy of the weak target and the traditional high order cumulations, as well as considering the spatial characteristics for enhancing the candidate targets in the prediction background. Finally, the diameter of dim targets was calculated by introducing the scale space theory to improve the pipeline filtering for fully filtering false targets and retaining the real dim targets. The test data show that this scheme can accurately detect the weak and small targets under the condition of noise and complex infrared background, which has a higher signal clutter gain and background suppression factor, as well as a better ROC curve.

关 键 词：红外弱小目标检测 各向异性背景抑制 时空高阶累积量 边缘停止函数 尺度空间 管道滤波 信杂比增益 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN21[自动化与计算机技术—计算机科学与技术]