鱼类行为与水体环境密切相关，是鱼类生活状况的直接体现，可以通过分析鱼类行为进行更为精准的养殖管理和操作。计算机视觉技术为鱼类行为识别和量化提供了一种非入侵式且稳定性较好的方法，已逐渐广泛用于鱼类行为研究。本文介绍了计算机视觉技术的技术流程包括图像采集、预处理、运动目标检测与跟踪，并对各个流程进行分类；综述了计算机视觉技术在鱼类游泳、摄食和体色变化等行为识别、量化研究的现状，分析了计算机视觉技术在鱼类行为识别、量化方面的难点及存在的问题，以期为计算机视觉技术在水产养殖监测领域的发展提供参考依据。

Fish behavior is closely related to the aquatic environment and is the direct embodiment of fish living conditions. Fish behavior can be analyzed for more accurate aquaculture management and operation. A non-invasive method with better stability for recognition and quantification of fish behavior is provided by computer vision technology, which has been widely used in fish behavior research. This paper introduces and classifies the processes of computer vision technology, including image acquisition, preprocessing, moving target detection and tracking, summarizes current status of computer vision technology in the research of recognition and quantification of fish behavior such as swimming, feeding and body color change, and analyzes the difficulties and existing problems of computer vision technology in recognition and quantification of fish behavior , hoping to provide reference for the development of computer vision technology in the field of aquaculture monitoring.