Multi-Channel Based on Attention Network for Infrared Small Target Detection

作     者：张彦军 王碧云 蔡云泽 ZHANG Yanjun;WANG Biyun;CAI Yunze

作者机构：Department of AutomationShanghai Jiao Tong UniversityShanghai 200240China Shanghai Institute of Aerospace Control TechnologyShanghai 201109China Infrared Detection Technology R&D Center of China Aerospace Science and Technology CorporationShanghai 201109China Key Laboratory of System Control and Information ProcessingMinistry of Education of ChinaShanghai 200240China Shanghai Engineering Research Center of Intelligent Control and ManagementShanghai 200240China Key Laboratory of Marine Intelligent Equipment and System of Ministry of EducationShanghai Jiao Tong UniversityShanghai 200240China Institute of Marine EquipmentShanghai Jiao Tong UniversityShanghai 200240China 

出 版 物：《Journal of Shanghai Jiaotong university(Science)》 (上海交通大学学报（英文版）)

年 卷 期：2024年第29卷第3期

页      面：414-427页

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：the Industry-University-Research Cooperation Fund Project of the Eighth Research Institute of China Aerospace Science and Technology Corporation (No.USCAST2021-5) the Major Scientific Instrument Research of National Natural Science Foundation of China (No.61627810) the National Science and Technology Major Program of China (No.2018YFB1305003) the National Defense Science and Technology Outstanding Youth Science Foundation (No.2017-JCJQ-ZQ-031) 

主　　题：infrared image small target detection deep learning attention mechanism feature fusion 

摘      要：Infrared detection technology has the advantages of all-weather detection and good concealment,which is widely used in long-distance target detection and tracking ***,the complex background,the strong noise,and the characteristics of small scale and weak intensity of targets bring great difficulties to the detection of infrared small targets.A multi-channel based on attention network is proposed in this paper,aimed at the problem of high missed detection rate and false alarm rate of traditional algorithms and the problem of large model,high complexity and poor detection performance of deep learning ***,given the difficulty in extracting the features of infrared multiscale and small dim targets,the multiple channels are designed based on dilated convolution to capture multiscale target ***,the coordinate attention block is incorporated in each channel to suppress background clutters adaptively and enhance target *** addition,the fusion of shallow detail features and deep abstract semantic features is realized by synthesizing the contextual attention fusion ***,it is verified that,compared with other state-of-the-art methods based on the datasets SIRST and MDFA,the proposed algorithm further improves the detection effect,and the model size and computational complexity are smaller.