红外图像小目标检测算法研究进展

Research Progress of Small Target Detection Algorithms in Infrared Images

红外小目标检测(Infrared Small Target Detection,ISTD)是远程精确打击、空天攻防对抗和遥感情报侦察等军事应用中的关键技术.红外图像由于目标尺寸小,缺乏颜色、形状信息,在复杂背景下对小目标检测一直是一项挑战性的问题.近年来,研究人员对ISTD开展了大量研究.根据研究方式和角度的不同,将小目标检测分为基于模型驱动的模型和数据驱动的模型两大类,并分别进行了简要梳理,分析了不同方法的原理、优势及不足.最后,指出了该领域未来的研究方向,即更关注性能高、泛化性好、复杂度低的检测算法.

Infrared small target detection(ISTD)is a key technology in military applications such as long-range precision strike,air and space attack and defense confrontation as well as remote sensing intelligence reconnaissance.Due to the small size of the target,infrared image lacks color and shape information,so the detection of small targets in complex background has been a very challenging problem.In recent years,researchers have carried out a lot of researches on ISTD.According to different perspectives of the research approach,small target detection is divided into two categories:model-driven and data-driven based models,both of which are briefly sorted out respectively.Moreover,the principles,advantages and shortcomings of different methods were analyzed.Finally,the future research direction of this field is pointed out,which is to pay more attention to the detection algorithms with high performance,good generalization and low complexity.