Evaluation of fabric pilling using hybrid imaging methods

A study has been made on the quantification and evaluation of fabric pilling using two-dimensional and three-dimensional hybrid imaging methods. Two-dimensional imaging method was good for some samples while three-dimensional measurement method for others, according to the properties of their base fabric. Various image processing techniques as well as three-dimensional data processing algorithms were applied for the extraction of pills from measured data and a series of shape parameters have been defined for the objective evaluation of fabric pilling. An evaluation criterion that is compatible with the conventional evaluation method has been proposed by applying the new evaluation method to the current photographic standards.     

Development of an objective fabric pilling evaluation method. I. Characterization of pilling using image analysis

An objective evaluation method for woven fabric pilling has been developed using image processing and statistical analysis. Five ASTM photographic standard images were analyzed to determine a characterization method for fabric pilling. Images were filtered by various image analysis techniques such as fast Fourier transformation (FFT) and fast wavelet transformation (FWT). Three parameters including the number of pills, the total pixel area of pilling, and the sum of the gray values of pill images have been defined and extracted automatically from the images. Finally, the relationship between pilling grade and those parameters has been established by a series of statistical analyses.     

Development of an objective fabric pilling evaluation method. II. Fabric pilling grading using artificial neural network

An objective pilling evaluation method has been developed using image analysis and artificial neural network. Pilling parameters obtained in the previous study were used as the input values for neural network. A total of 9 data sets including 5 standard grades and 4 interpolated intermediate grades were used for training the network. Nine samples were prepared to verify the validity of the trained network in comparison with the subjective evaluation results. 18 woven and 12 knitted samples were tested to investigate the effect of the fabric structure on the performance of the network. Finally, 55 woven fabric samples were tested to evaluate the performance of the newly developed method and it was proven to be suitable for the evaluation of pilling grade especially for woven fabrics.     

Image inspection of nonwoven defects using wavelet transforms and neural networks

Image inspection of nine kinds of nonwoven defects by the wavelet transform and neural network is presented. The defects include black yarn, hole, needle streak, oil stain, stripe, corrugation, white spot, folding mark, and wrinkle mark. The wavelet transform decomposes an original image into four subimages in different frequency bands. Four texture measures, the energy, contrast and correlation with gray-level co-occurrence matrices as well as the energy with wavelet coefficients, are selected as defect features and computed based on the low-frequency subimage at resolution level one. The feature values are distributed in groups by the categories of defects throughout the feature space, accounting for suitability of the four features for inspecting nonwoven defects. The subimage is a downsized approximation of the original image; thus, in this manner, feature extraction can not only consume less computation time but also maintain the classification accuracy. The neural network acts as a classifier, which is trained by forty-five samples. The experimental results demonstrate that among forty-five testing samples, the classification accuracy is 100 %.     

Intelligence control of on-line dynamic gray cloth inspecting machine system module design. II. Defects inspecting module design

This study aimed to establish a set of gray cloth defect inspection module using image processing technique, so as to develop a full intelligent online dynamic gray cloth defect automatic inspection system. Gray cloth defects to be recognized in this study included holes, stains, warp missing, spider web and weft missing. First use wavelet transform and co-occurrence matrix to find features of gray cloth defect image, next, use back-propagation neural network (BPNN) to make gray cloth defect classification and data output. BPNN was capable of solving nonlinear problems, thus assisted in enhancing defect recognition effect. As every defect to be inspected in this study varied in size and shape, so advantage of BPNN could be used as aid more than else. This study primarily utilized image processing technique to inspect gray cloth defects, not only in a faster speed than common visual inspection, but also eliminating arbitrary factors of inspectors in body and psychology during inspection, resulting in absolute objectivity. Finally, tension control module built in Part 1 and gray cloth defect inspection module built in this study were integrated, and a full intelligent online dynamic gray cloth defect automatic inspection system established. As validated by experiment result, the system established in this study could successfully recognize gray cloth defects, with total recognition rate amounting to 92.5 %.     

The evaluation of evenness of nonwovens using image analysis method

Authors studied on the applicability of image analysis technique using a scanner with a CCD (charged coupled deviced) to the evaluation of evenness of nonwovens because it has distinctive features to considerably save time and labor in the analysis compared with other classical methods. As specimens for the experiment, two different types that are unpatterned and patterned ones were prepared. For the unpatterned specimen, webs were chermically bonded, while for the patterned specimen, webs being thermally calendered with engraved roller. Several webs having various areal densities were prepared and bonded. Coefficient of variation (CV%) was used as a parameter to evaluate the evenness. Scanning conditions could be suitably set up through comparing the total variance to the between-group variance and to the within-group variance, respectively, on the images scanned at the different conditions. The 2D convolution method with smoothing filter kernel was introduced to further filter the noises on the scanned images. After the filtering process, the increase of web areal densities gave an uniform decrease of the CV%. This showed that the scanned image analysis with proper filtering process could be successfully applicable to the evaluation of evenness in nonwovens.