In this study, a new detection algorithm for yarn-dyed fabric defect based on autocorrelation function and grey level co-occurrence matrix (GLCM) is put forward. First, autocorrelation function is used to determine the pattern period of yarn-dyed fabric and according to this, the size of detection window can be obtained. Second, GLCMs are calculated with the specified parameters to characterise the original image. Third, Euclidean distances of GLCMs between being detected images and template image, which is selected from the defect-free fabric, are computed and then the threshold value is given to realise the defect detection. Experimental results show that the algorithm proposed in this study can achieve accurate detection of common defects of yarn-dyed fabric, such as the wrong weft, weft crackiness, stretched warp, oil stain and holes.

Due to the significant potential values in commercial and social applications, clothing image recognition has recently become a research hotspot, among which clothing attribute recognition is an important content. However, the large variations in the appearance and style of clothing and the image’s complex forming conditions make the task challenging. Moreover, a generic treatment with deep convolutional neural networks cannot provide an ideal solution. Instead of using CNNs for classification, we proposed a novel approach based RCNN framework for the recognition task. Firstly, we apply the modified selective search algorithm to extract the region proposal. Then, the Inception-ResNet V1 model with L-Softmax is employed to represent images and identify their categories. After Soft-NMS, we use a simple neural network to correct the boundary of region box. To evaluate the performance of the framework, a dataset including about 100,000 shirt images was built. The experimental result show that our proposed framework achieved promising overall labeling rate, precision and recall of 87.77%, 73.59% and 83.84%. In addition, comparative experiments demonstrate the superiority of the proposed framework.

Yarn density measurement is a significant part of yarn-dyed fabric analysis, traditionally based on reflective image analysis. In this paper, utilizing fabric light transmittance, a method for two-dimensional discrete Fourier transform (2D DFT) analysis on the transmission fabric image is developed for fabric density inspection. First, the power spectrum is generated from the fabric image by a 2D DFT. Next, the yarn skew angles are detected based on the power spectrum analysis. Then the fabric image is reconstructed by an inverse 2D DFT. Finally, projection curves are generated from the reconstructed images and the number of yarns is counted according to the peaks and valleys to obtain the fabric density. Through a comparison between analysis on the reflective and transmission images of multiple-color fabrics, it is proved that the latter method can segment the yarns with more satisfactory accuracy. Furthermore, the experimental and theoretical analyses demonstrate that the proposed method is effective for the density inspection of yarn-dyed fabrics with good robustness and great accuracy.

Due to the potential values in many areas such as e-commerce and inventory management, fabric image retrieval, which is a special case in Content Based Image Retrieval (CBIR), has recently become a research hotspot. It is also a challenging issue with serval obstacles: variety and complexity of fabric appearance, high requirements for retrieval accuracy. To address this issue, this paper proposes a novel approach for fabric image retrieval based on multi-task learning and deep hashing. According to the cognitive system of fabric, a multi-classification-task learning model with uncertainty loss and constraint is presented to learn fabric image representation. Then we adopt an unsupervised deep network to encode the extracted features into 128-bits hashing codes. Further, the hashing codes are regarded as the index of fabrics image for image retrieval. To evaluate the proposed approach, we expanded and upgraded the dataset WFID, which was built in our previous research specifically for fabric image retrieval. The experimental results show that the proposed approach outperforms the state-of-the-art.

This article presents a novel method for measuring the unevenness of yarn apparent diameter based on yarn sequence images captured from a moving yarn. A dynamic threshold module was designed to gain the global threshold for segmenting yarns in the sequence images. In the module, a K -means clustering algorithm was employed to classify the pixels of each frame in the sequence into two clusters—yarn and background. The cluster center of the current frame was used as the initial value of the cluster center for the next frame in the sequence to expedite the segmentation process. From the segmented yarn image, the yarn core was further extracted utilizing the characteristics of yarn hairiness, and two judgment templates were adopted to remove burrs, isolated points and unrelated small areas in the images. The yarn apparent diameter was measured on the yarn core at a given interval. The same kind of yarns were tested by using this method and Uster Evenness Tester 5. The experimental results show that the proposed method can accurately detect the unevenness of yarn apparent diameter and provide new useful information about yarn unevenness, such as the short-term, the long-term, and the periodic variations of yarn apparent diameters.