Rapid variety identification of pure rough rice by Fourier-transform near-infrared spectroscopy.

摘要

Rice variety is considered as an important factor influencing cooking and processing quality because of variations in size, shape and constitution. Difficulty in the management of rough rice with lower varietal purity becomes a significant problem in rice production and can result in the reduction of rice quality. Fourier-transform near-infrared (FT-NIR) spectroscopy was used to identify the variety of rough rice through wholegrain techniques. Moist rough rice samples (n = 259) comprising 5 varieties (Khao Dawk Mali 105 (KDML105), Pathum Thani 1, Suphan Buri 60, Chainat 1 and Pitsanulok 2) were gathered from different locations around Thailand and scanned in the NIR region of 9088-4000 cm--1 in reflectance mode. Soft independent modelling of class analogies (SIMCA) and partial least squares discriminant analysis (PLSDA) methods were used for identification by utilizing preprocessed spectra. The highest identification accuracy achieved was 74.42% by the SIMCA model and 99.22% by the PLSDA model. The best PLSDA model demonstrated approx. 97% correct identification for KDML105 samples and 100% for the others. It is suggested that this study may raise the possibility of applying FT-NIR spectroscopy as a nondestructive technique for rapidly identifying moist rough rice varieties in routine quality assurance testing.