Evaluating Fabric Pilling with Light-Projected Image Analysis

Abstract

An innovative method and a device for objectively evaluating fabric pilling grade based on light projection using image analysis is presented in this paper. The method can eliminate interference with pilling information from fabric color and pattern. The device for acquiring light-projected images, detecting the profile of projected images, segmenting pills appearing on converted gray images, extracting of a pill's feature index, and so on are discussed in detail. Pilling grade is assessed by a Kohonen self-organizing feature map neural network. Thirty different kind of pilled fabric samples are trained and tested, and the correlation coefficient between the objective grade and subjective grade is 0.94 for the training samples and 1 for the testing samples. The sample number with ± I grade deviation is 5, so the objective inspection accuracy is 83%.Pilling on textile fabrics is a well-known phenomenon.The development of pills on a fabric surface, in addition to spoiling the original appearance and hand, initiates garment attrition and reduces serviceability. Pilling resistance is usually tested in the textile trade by quality control, where the fabric sample is tumbled or treated with abrasive materials to simulate normal wear. Then the pilled sample is compared with standard photographs of pilled fabric samples by visual inspection to evaluate the pilling degree that usually ranges from grade 5 (no pilling) to grade 1 (very severe pilling). Because of the inconsistency and inaccuracy of rating results obtained with the visual method, more reliable and objective methods for pilling evaluation are desirable for the textile industry.Many textile scientists have made valuable contributions to objective assessments of fabric piling [1][2][3][4][5][8][9][10][11][12][13]]. All these works, however, face the common difficulty of interference with fabric pill information from fabric color and pattern, which eventually affects pill ratings. In this paper, we propose a new method of objectively assessing fabric pilling based on light-projected image analysis, which may be the way to overcome this problem. > Experimental Device for Acquiring Light-Projected Images Figure 1 shows the device we have designed and installed to acquire the projected cross-sectional images of pilled fabric samples. Samples ( 1 ) are fixed in advance on a conveyer belt (2) driven by a motor (3) and can be wrapped closely around a shaped bar (4), which is specially designed to have a triangular cross-sectional shape and to be characteristic of a large curvature in the bar edge. When-the sample arrives at the edge of the bar where a light source (5) is positioned below, sequential projected cross-sectional images of the pilled fabric samples are captured by a CCD camera (7) positioned above the bar. Those projected images are then read into the computer to be analyzed.

Adaptive neuro-fuzzy system for quantitative evaluation of woven fabrics’ pilling resistance

“…This process can be done using a digital scanner (Kim & Kang, 2005;Palmer, Joud, & Wang, 2005;Palmer & Wang, 2003;Semnani & Ghayoor, 2009;Yun, Kim, & Park, 2013), a camera (Behera & Mohan, 2005;Hsi, Bresee, & Annis, 1998a;Torres & Navarro, 1998;Xin, Hu, & Yan, 2002), a light projected on camera (Chen & Huang, 2004), a camera attached to a microscope (Izabela, 2009), optical triangulation topographic reconstruction of the fabric surface (De Oliveira Mendes, Fiadeiro, & Miguel, 2010, 2011De Oliveira Mendes, Fiadeiro, Miguel, & Lucas, 2009), a laser line projected on the surface of the fabric specimen (Kang, Cho, & Kim, 2004), or a stereovision surface reconstruction using two CCD cameras (Kang et al, 2004).…”

Artificial Neural Network Prosperities in Textile Applications

“…Because of the inconsistency and inaccuracy of rating results obtained with the visual method, more reliable and objective methods for pilling evaluation are desirable for the textile industry. Chen & Huang, 2004 evaluated and graded fabric pilling based on light projection using image analysis and neural network to overcome the common difficulty of interference with fabric pill information from fabric color and pattern. Firstly, they eliminated interference with pilling information from fabric color and pattern.…”

“…The total number of iterations in the training process was 400, and the learning rate was initialized to be 0.02.They concluded that the objective pilling grade was in good agreement with the subjective pilling grade. The correlation coefficient for training and testing samples were reported up to 0.94 and 1 respectively (Chen & Huang, 2004). Beltran et al, 2005 also used artificial neural networks to model the multi-linear relationship between fiber, yarn and fabric properties and their effect on the pilling propensity of pure wool knitted fabrics.…”

Artificial Neural Networks and Their Applications in the Engineering of Fabrics

“…Further on, the impact permeability has been studied (Tokarska & Gniotek, 2009) and the quality of the neural models has been assessed (Tokarska, 2006). The pilling propensity of the fabrics has been predicted (Beltran et al, 2005) and the pilling of the fabrics has been evaluated (Chen & Huang, 2004;, while the presence of fuzz fibres has been modelled (Ucar & Ertugrul, 2007). The evaluation of the wrinkle of the fabrics has been realized on an objective basis with a system based on ANNs (Su & Xu, 1999;Kim, 1999;Mori & Komiyama, 2002).…”