Near infrared spectroscopy combined with chemometrics analysis were investigated as an emerging method for the identification of textiles, this method allowed straightforward and rapid testing of textiles without destroying their integrity compared to traditional testing methods such as dissolution method, combustion method and microscope observation. In this work, different pretreating algorithms coupled with the Soft Independent Modelling by Class Analogy (SIMCA) have been studied to achieve the best recognition model to identify different pure yarn fabrics (cotton, linen, silk, wool and polyester). Results showed that little difference between different data points smoothing. Percentages of recognition and rejection of 100% were obtained of silk, wool and polyester by pretreating with Savitzky-Golay 7 data points smoothing, Savitzky-Golay second derivative and mean centering. The percentages of recognition were 93.75% and 96.25% for cotton and linen respectively by pretreating with Savitzky-Golay 7 data points smoothing, Savitzky-Golay first derivative and mean centering, nevertheless, the percentages of rejection was low for linen with 83.75%. The results from this paper suggested that near infrared spectroscopy in combination with SIMCA could be applied to the identification of pure yarn fabrics, whereas further study should be made to improve the rejection rate of linen.
Use of Near-Infrared Spectroscopy and Chemometrics for the Nondestructive Identification of Concealed Damage in Raw Almonds (Prunus dulcis)