红外图像边缘检测算法综述

为填补红外图像边缘检测算法综述性研究的空白，使更多研究者较为全面地了解目前成果，并为后续研究提供有价值的参考，遴选了近十年国内外红外图像边缘检测技术研究的相关文献。首先概述了红外成像与边缘检测技术，进而阐述了红外图像边缘检测技术的难点与挑战，接着总结了主要的红外图像边缘检测算法，将相关算法分为了4类——基于经典边缘检测算子改进的、基于蚁群算法的、基于数学形态学的和基于网络模型的，对其涉及的关键技术分别进行了分析。研究认为，在传统红外图像边缘检测技术中，形态学方法因简单易用而具有一定潜力；对于非传统红外图像边缘检测技术，基于深度学习的方法对目标边缘的针对性更强、鲁棒性更好、不需要设计复杂的算法步骤，给红外图像边缘检测带来了新的发展机遇。

Review of Infrared Image Edge Detection Algorithms

To ensure that researchers are well-informed regarding infrared image edge detection algorithms and to provide a valuable reference for follow-up investigations,we review relevant research conducted on infrared image edge detection algorithms in the past ten years.First,infrared imaging and edge detection technology are summarized,and then,the difficulties and challenges of infrared image edge detection algorithms are described.Finally,the main infrared image edge detection algorithms are summarized,and the related algorithms are divided into four categories:improved classic edge detection operator-based algorithms,ant colony algorithm-based algorithms,mathematical morphology-based algorithms,and network model-based algorithms.Considering traditional infrared image edge detection algorithms,the morphological method has potential because of its simplicity and ease of use;for non-traditional infrared image edge detection algorithms,the method based on deep learning has stronger pertinence,better robustness,and no requirement of designing complex algorithm steps,which brings new development opportunities to infrared image edge detection.