低虚警率的红外弱小目标检测算法  

An Infrared Dim Target Detection Algorithm with Low False Alarm Rate

作　　者：陈紫强[1] 梁晨 CHEN Ziqiang;LIANG Chen(Department of Information and Communication Engineering,Guilin University of Electronic Technology,Guilin 541004,China;Ministry of Education Key Laboratory of Cognitive Radio and Information Processing,Guilin University of Electronic Technology,Guilin 541004,China)

机构地区：[1]桂林电子科技大学信息与通信工程系,广西桂林541004 [2]桂林电子科技大学认知无线电与信息处理教育部重点实验室,广西桂林541004

出　　处：《电讯技术》2022年第11期1600-1606,共7页Telecommunication Engineering

基　　金：国家自然科学基金资助项目(61861011);广西研究生教育创新计划项目。

摘　　要：针对远距离复杂场景下红外弱小目标信噪比低导致目标检测虚警率高的问题,提出了一种时域与空域滤波相融合的红外弱小目标检测方法。采用相对局部对比度算法(Relative Local Contrast Measure,RLCM)增强目标信噪比,抑制高亮度背景;利用目标的时空相关性,运用时域局部差分算法(Temporal Local Difference Algorithm,TLCD)增强目标,消除固定噪点。融合空域和时域的检测结果获得时空相对局部对比度图(Spatial-Temporal Relative Local Contrast Map,STRLCM),通过自适应阈值分割提取待检测的真实目标。实验结果表明,与现有算法相比,所提算法可以极大地降低虚警率同时保持较高的检测效果。For the problem that the low signal-to-noise ratio(SNR)of infrared dim and small targets leads to high false alarm rate in target detection in long-distance complex scenes,an infrared small target detection method combining time domain and spatial filtering is proposed.The relative local contrast measure(RLCM)is used to enhance the SNR of the target and suppress the high-brightness background.By utilizing the temporal and spatial correlation of target,the temporal local difference algorithm(TLCD)is used to enhance the target and eliminate the fixed noise.The spatial-temporal relative local contrast map(STRLCM)is obtained based on detection results in both spatial and temporal domains,and the to-be-detected real target is extracted via the adaptive threshold segmentation.Experimental results show that compared with the existing algorithms,the proposed algorithm can greatly reduce the false alarm rate while maintaining a high detection effect.

关 键 词：红外弱小目标检测 远距离复杂场景 相对局部对比度(RLCM) 时域局部差分(TLCD) 时空相对局部对比度 自适应阈值分割 

分 类 号：TN911.73[电子电信—通信与信息系统]