Kuo Chung-Feng Jeffrey, Wang Wei-Ren, Barman Jagadish
Department of Materials Science and Engineering, National Taiwan University of Science and Technology, Taipei 10607, Taiwan.
Sensors (Basel). 2022 Sep 24;22(19):7246. doi: 10.3390/s22197246.
This paper presents a turnkey integrated system that can be operated in real time for real textile manufacturers. Eight types of defects in woven fabric, including stain, broken end, broken weft, hole, nep, double pick, kinky weft and float can be recognized and classified. First, an image is captured by a CMOS industrial camera with a pixel size of 4600 × 600 above the batcher at 20 m/min. After that, the four-stage image processing procedure is applied to detect defects and for classification. Stage 1 is image pre-processing; the filtration of the image noise is carried out by a Gaussian filter. The light source is corrected to reduce the uneven brightness resulting from halo formation. The improved mask dodging algorithm is used to reduce the standard deviation of the corrected original image. Afterwards, the background texture is filtered by an averaging filter, and the mean value is corrected for histogram shifting, so that this system is robust to the texture and color changes of woven fabric. The binary segmentation threshold is determined using the mean value and standard deviation of an image with a normal sample. Stage 2 uses adaptive binarization for separation of the background and defects and to filter the noise. In Stage 3, the morphological processing is used before the defect contour is circled, i.e., four features of each block, including the defect area, the aspect ratio of the defect, the average gray level of the defect and the defect orientation, which are calculated according to the range of contour. The image defect recognition dataset consists of 2246 images. The results show that the detection success rate is 96.44%, and the false alarm rate is 3.21%. In Stage 4, the defect classification is implemented. The support vector machine (SVM) is used for classification, 230 defect images are used as training samples, and 206 are used as test samples. The experimental results show that the overall defect recognition rate is 96.60%, providing that the software and hardware equipment designed in this study can implement defect detection and classification for woven fabric effectively.
本文介绍了一种可实时运行的交钥匙集成系统，供实际的纺织制造商使用。该系统能够识别和分类机织织物中的八种缺陷，包括污渍、断经、断纬、破洞、棉结、双纬、扭纬和浮纬。首先，在20米/分钟的速度下，通过一台像素尺寸为4600×600的CMOS工业相机在配料机上方捕获图像。之后，应用四阶段图像处理程序来检测缺陷并进行分类。第一阶段是图像预处理；通过高斯滤波器对图像噪声进行过滤。对光源进行校正，以减少光晕形成导致的亮度不均匀。使用改进的蒙版减淡算法来降低校正后的原始图像的标准差。然后，通过均值滤波器对背景纹理进行过滤，并对均值进行校正以进行直方图平移，从而使该系统对机织织物的纹理和颜色变化具有鲁棒性。使用正常样本图像的均值和标准差确定二值分割阈值。第二阶段使用自适应二值化来分离背景和缺陷并过滤噪声。在第三阶段，在圈出缺陷轮廓之前进行形态学处理，即根据轮廓范围计算每个块的四个特征，包括缺陷面积、缺陷的长宽比、缺陷的平均灰度级和缺陷方向。图像缺陷识别数据集由2246张图像组成。结果表明，检测成功率为96.44%，误报率为3.21%。在第四阶段，进行缺陷分类。使用支持向量机（SVM）进行分类，将230张缺陷图像用作训练样本，206张用作测试样本。实验结果表明，总体缺陷识别率为96.60%，这表明本研究设计的软硬件设备能够有效地实现机织织物的缺陷检测和分类。
