基于深度学习的红外小目标检测算法综述

A review of infrared small target detection algorithms based on deep learning

红外检测技术具有受环境负面影响小、抗外界干扰能力强等优势,在众多领域皆有极为重要的应用价值。然而,由于红外小目标存在缺少明显的可用信息、边界模糊等问题,对其检测的难度较大,因而成为目标检测领域的研究热点与难点。本文通过分析困扰红外小目标检测研究发展的难题所在,首先就目前针对其检测的传统算法原理进行简要说明。其次,详细阐述了基于深度学习的多类型红外小目标检测算法,并对相关算法的分类、评估指标、相关数据集等多方面内容进行了介绍,随之以实例说明对当前算法改进的有效方式。最后,归纳总结现有检测算法的优缺点,探讨了红外小目标检测研究领域的未来发展趋势,即向高精度、高实时性、强鲁棒性、低复杂度的算法方面深入研究。

Infrared detection technology has the advantages of small negative impact by the environment,strong resistance to external interference,etc.,and has extremely important application value in many fields.However,small infrared targets are difficult to detect due to the lack of apparently available information and blurred boundaries and other problems,thus becoming a research hotspot and difficulty in the field of target detection.In this paper,by analyzing the problems of infrared small target detection research and development,firstly,the principles of traditional algorithms for its detection are briefly explained;Secondly,the multi type infrared small target detection algorithm based on deep learning is described in detail,and the classification,evaluation indicators,relevant data sets and other aspects of the relevant algorithms are introduced,and then the effective way to improve the current algorithm is illustrated by examples.Finally,the advantages and disadvantages of the existing detection algorithms are summarized,and the future trends in the research field of infrared small target detection research is discussed,that is,the algorithms with high accuracy,high real time performance,strong robustness and low complexity are further studied.