Automated recognition and sorting of recycled textiles for sustainable fashion

Abstract

The application of the principles of sustainable fashion is one of the solutions to reduce the amount of waste from textile production and the use of such fabrics. Spectrophotometric methods have effective application in this subject area. In the present work, an analysis of known methods and approaches applied so far using the techniques of spectral analysis. The proposed methods and procedures lead to improvement and facilitation of the process of classification of textile fibers in sorting and recycling of textile fabrics, in order to implement in automated systems. The proposed analysis tools do not require high cost equipment and complex calculation procedures. They can be implemented in portable devices and microprocessor-based recognition systems. It has been found that two principal components and two latent variables are sufficient to describe the variance in the data. This significantly reduces the amount of data used to analyze textile fibers by their spectral characteristics. It has been shown that the accuracy of textile fiber classification does not depend on the type of separation function of the classifier used. This accuracy depends on the spectral characteristics used, the method for reducing the volume of data, and the type of classifier. The obtained results can be used in the development of recognition systems for sorting textile fabrics depending on the composition of their fibers. In this way, the principles of sustainable fashion will be effectively applied. Also, the proposed methods and tools can be used in the training of future specialists in the subject area.