Abstract: With the wide application and large-scale implementation of machine vision in various fields, the application of machine vision in textile production and quality control has also become a research hotspot. Firstly, the hardware selection and arrangement of the fabric defect detection machine vision detection system are introduced, and the selection of the light source and the camera is mainly introduced. Then according to the different detection principles, the machine vision detection methods are divided into statistical methods, spectrum methods, learning-based methods, structural methods and model methods, and an overview of these methods is done. Finally, the paper analyzes the problems faced by machine vision in fabric defect detection and the future development concept.

Key words: machine vision, image decomposition, fabric defect detection, automatic recognition

摘要： 随着机器视觉在各个领域的广泛应用和大面积实施，关于机器视觉在纺织生产和质量控制上的应用也成为了研究热点。介绍了织物疵点机器视觉检测系统的硬件选择与布置，重点介绍了光源和相机的选取。根据检测原理的不同，将机器视觉检测方法分为统计法、频谱法、基于学习的方法以及结构法和模型法，并对这些方法做了综述。梳理分析了机器视觉在织物疵点检测方面所面临的问题和未来发展构想。

关键词: 机器视觉, 图像分解, 织物疵点检测, 自动识别