联合γ-范数和TV-稀疏约束的红外弱小目标检测  

Joint Constraint Based onγ-Norm and TV-Sparse for Infrared Dim Small Target Detection

作　　者：王孝文 李乔 薛伟 钟平[2] Wang Xiaowen;Li Qiao;Xue Wei;Zhong Ping(School of Computer Science and Technology,Anhui University of Technology,Maanshan 243032,China;National Key Laboratory of Science and Technology on ATR,National University of Defense Technology,Changsha 410073,China)

机构地区：[1]安徽工业大学计算机科学与技术学院,安徽马鞍山243032 [2]国防科技大学ATR重点实验室,长沙410073

出　　处：《航空兵器》2022年第2期30-38,共9页Aero Weaponry

基　　金：中国博士后科学基金特别项目(2020T130767);自动目标识别重点实验室稳定支持项目(WDZC20195500206)。

摘　　要：针对基于传统块图像模型的红外弱小目标检测算法对背景杂波抑制能力不强的问题,提出了一种联合γ-范数和全变分正则化与稀疏约束建模的红外弱小目标检测模型(γ-TSIPI)。首先,将原始红外图像转化为红外块图像,然后,采用γ-范数和全变分正则化对背景块图像进行约束,以更好地减少目标图像中的残留噪声,同时保留图像的边缘信息,避免恢复的背景图像过度光滑。此外,考虑到传统红外块图像模型中的L_(1)范数会过度缩小弱小目标,引入了加权的L_(1)范数,以提升γ-TSIPI模型对目标图像的恢复能力。最后,应用Lagrange乘子法求解γ-TSIPI模型。实验结果表明,所提方法可以更好地抑制背景杂波,降低虚警率,有效地提高了检测性能。Aiming at the problem that the ability of infrared dim small target detection algorithm based on traditional infrared patchimage(IPI)model to suppress background clutter is not strong,a new detection model(γ-TSIPI)based onγ-norm,total variational regularization,and sparse constraint modeling is proposed.Firstly,the original infrared image is transformed into an IPI,and then theγ-norm and total variational regularization are used to constrain the background patch image to reduce the residual noise in the target image.At the same time,the edge information of the image is retained to avoid excessive smoothness of the restored background image.In addition,considering that the L_(1) norm in the traditional IPI model may reduce the dim small target excessively,the weighted L_(1) norm is introduced to improve the recovery ability ofγ-TSIPI model.Finally,the Lagrange multiplier method is applied to solve theγ-TSIPI model.Experimental results show that the proposed method can suppress background clutter better,reduce false alarm rate and effectively improve detection performance.

关 键 词：红外 弱小目标检测 红外块图像模型 单帧图像 γ-范数 全变分 稀疏 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程] TN911.73[电子电信—通信与信息系统]