Detection of the fluff fabric surface quality based on machine vision

• Published in: Journal of the Textile Institute Vol. 113; no. 8; pp. 1666 - 1676
• Main Authors: Lin, Qiangqiang, Zhou, Jinzhu, Ma, Qiurui
• Format: Journal Article
• Language: English
• Published: Manchester Taylor & Francis 25.07.2022; Taylor & Francis Ltd
• Subjects: Datasets; Decomposition; Fluff fabric; Image acquisition; Machine vision; neural network; Neural networks; optical imaging; Production methods; Quantitative analysis; Support vector machines; Surface properties; SVM
• ISSN: 0040-5000; 1754-2340
• DOI: 10.1080/00405000.2021.1943946

Fabric surface quality detection plays a vital role in the fabric production, and current methods are mainly based on the detection of fabric defects and pilling. In this paper, qualitative and quantitative evaluation models have been proposed for the evaluation of fluff fabric. Firstly, an image acquisition system was constructed according to the raising process, the region of interest (ROI) for the image was cropped by the horizontal gray projection. On the qualitative evaluation model, seven different neural network models were trained through two kinds of datasets, which were consisted of the fabric edge coordinates and the high-frequency information of the wavelet decomposition. On the quantitative evaluation model, we established the thickness parameter (Ra) and spacing model (Rs), which apply the upper edge coordinate dataset. The results showed that the accuracy of Support Vector Machine (SVM) model trained by using the coordinate dataset of wavelet decomposition was 99.42%, which is highest among other models, and quantitative evaluation results were consistent with the manual evaluation.