改进非凸估计与非对称时空正则化的红外小目标检测方法
2024,46(3):180-194
胡亮
中南大学 自动化学院, 湖南 长沙 410083，hliang0611@163.com
杨德贵
中南大学 自动化学院, 湖南 长沙 410083
赵党军
中南大学 自动化学院, 湖南 长沙 410083
张俊超
中南大学 自动化学院, 湖南 长沙 410083
中南大学 自动化学院, 湖南 长沙 410083，hliang0611@163.com
杨德贵
中南大学 自动化学院, 湖南 长沙 410083
赵党军
中南大学 自动化学院, 湖南 长沙 410083
张俊超
中南大学 自动化学院, 湖南 长沙 410083
摘要：
针对复杂背景下的红外小目标检测,在非对称时空正则化约束的非凸张量低秩估计算法基础上,提出了一种新的核范数估计方法代替原算法中的估计方法。提出基于结构张量与多结构元顶帽(Top-Hat)滤波的自适应权重张量对目标张量进行约束,增强目标张量稀疏性的同时抑制其中残存的强边缘结构。实验结果表明,所提改进算法能够更好地消除图像中强边缘结构对检测结果的影响,在保证检测率的情况下,较原算法具有更低的虚警率。
针对复杂背景下的红外小目标检测,在非对称时空正则化约束的非凸张量低秩估计算法基础上,提出了一种新的核范数估计方法代替原算法中的估计方法。提出基于结构张量与多结构元顶帽(Top-Hat)滤波的自适应权重张量对目标张量进行约束,增强目标张量稀疏性的同时抑制其中残存的强边缘结构。实验结果表明,所提改进算法能够更好地消除图像中强边缘结构对检测结果的影响,在保证检测率的情况下,较原算法具有更低的虚警率。
基金项目：
国家自然科学基金资助项目(62171475,62105372)
国家自然科学基金资助项目(62171475,62105372)
Infrared small target detection method based on improved non-convex estimation and asymmetric spatial-temporal regularization
HU Liang
School of Automation, Central South University, Changsha 410083, China,hliang0611@163.com
YANG Degui
School of Automation, Central South University, Changsha 410083, China
ZHAO Dangjun
School of Automation, Central South University, Changsha 410083, China
ZHANG Junchao
School of Automation, Central South University, Changsha 410083, China
School of Automation, Central South University, Changsha 410083, China,hliang0611@163.com
YANG Degui
School of Automation, Central South University, Changsha 410083, China
ZHAO Dangjun
School of Automation, Central South University, Changsha 410083, China
ZHANG Junchao
School of Automation, Central South University, Changsha 410083, China
Abstract：
Aiming at infrared dim and small targets detection in complex background, a new kernel norm estimation method was proposed based on the non-convex tensor low-rank approximation algorithm with asymmetric spatial-temporal total variation regularization, replacing the original estimation method in the algorithm. An adaptive weight tensor based on structure tensor and multi-structure element Top-Hat filtering was proposed to constrain the target tensor, which had enhanced the sparsity and suppressed the remaining strong edge structures of the target tensor. Experimental results show that the proposed improved algorithm can better eliminate the influence of strong edge structure on the detection results, and has a lower false alarm rate than the original algorithm under the condition of ensuring the detection rate.
Aiming at infrared dim and small targets detection in complex background, a new kernel norm estimation method was proposed based on the non-convex tensor low-rank approximation algorithm with asymmetric spatial-temporal total variation regularization, replacing the original estimation method in the algorithm. An adaptive weight tensor based on structure tensor and multi-structure element Top-Hat filtering was proposed to constrain the target tensor, which had enhanced the sparsity and suppressed the remaining strong edge structures of the target tensor. Experimental results show that the proposed improved algorithm can better eliminate the influence of strong edge structure on the detection results, and has a lower false alarm rate than the original algorithm under the condition of ensuring the detection rate.
收稿日期：
2022-05-19
2022-05-19
