Classification of textile fabrics by use of spectroscopy-based pattern recognition methods

The combination of near-infrared spectroscopy and pattern recognition methods, including soft independent modeling of class analogy, least squares support machine, and extreme learning machine, was employed for textile fabrics classification. The fabrics of cotton, viscose, acrylic, polyamide, polyester, and blend fabric of cotton-viscose were divided into training and prediction sets (60: 60) for developing models and evaluating the classification abilities of the models. The classification accuracy and speed of soft independent modeling of class analogy, least squares support machine, and extreme learning machine were compared. Both least squares support machine and extreme learning machine achieved the classification accuracy of 100% for the prediction set. However, extreme learning machine performed much faster than least squares support machine, which suggested that extreme learning machine may be a promising method for real-time textile fabrics classification with a comparable accuracy based on near-infrared spectroscopy. Moreover, it might have commercial and regulatory potential to avoid time-consuming work, and costly and laborious chemical analysis for textile fabrics classification.