Defects in the textile manufacturing process lead to a great waste of resources and further affect the quality of textile products. Automated quality guarantee of textile fabric materials is one of the most important and demanding computer vision tasks in textile smart manufacturing. This survey presents a thorough overview of algorithms for fabric defect detection. First, this review briefly introduces the importance and inevitability of fabric defect detection towards the era of manufacturing of artificial intelligence. Second, defect detection methods are categorized into traditional algorithms and learning-based algorithms, and traditional algorithms are further categorized into statistical, structural, spectral, and model-based algorithms. The learning-based algorithms are further divided into conventional machine learning algorithms and deep learning algorithms which are very popular recently. A systematic literature review on these methods is present. Thirdly, the deployments of fabric defect detection algorithms are discussed in this study. This paper provides a reference for researchers and engineers on fabric defect detection in textile manufacturing.

Ensemble classifiers improve the classification accuracy by incorporating the decisions made by its component classifiers. Basically, there are two steps to create an ensemble classifier: one is to generate base classifiers and the other is to align the base classifiers to achieve maximum accuracy integrally. One of the major problems in creating ensemble classifiers is the classification accuracy and diversity of the component classifiers. In this paper, we propose an ensemble classifier generating algorithm to improve the accuracy of an ensemble classification and to maximize the diversity of its component classifiers. In this algorithm, information entropy is introduced to measure the diversity of component classifiers, and a cyclic iterative optimization selection tactic is applied to select component classifiers from base classifiers, in which the number of component classifiers is dynamically adjusted to minimize system cost. It is demonstrated that our method has an obvious lower memory cost with higher classification accuracy compared with existing classifier methods.