Study on Rapid and Non-Destructive Identification of Starch Categories with NIR (Near-Infrared Reflectance)
We conduct rapid and non-destructive experimental studies on various starches based on NIR and cluster analysis (a kind of quantitative analysis), taking a total of 54 samples from two starches as subjects, and study the feasibility of identifying starch categories with the combination of NIR and cluster analysis. The samples are identified quantitatively through cluster analysis (a kind of spectrum pattern recognition) under NIRDRS (near-infrared diffuse reflectance spectroscopy) with spectrum from 8914-1 to 4000-1 cm. Researches indicate that once the spectra are pre-processed trough vector normalization, samples can be classified into two categories accurately with cluster analysis, with a precision rate of 100%.In the meantime, the accuracy rate from experiment is also 100% when using the prediction set samples to test the accuracy of the model built. As research shows, a combination of NIRDRS and cluster analysis provides an accurate and reliable method to recognize starch categories rapidly and non-destructively.
X. R. Sun et al., "Study on Rapid and Non-Destructive Identification of Starch Categories with NIR (Near-Infrared Reflectance)", Advanced Materials Research, Vols. 383-390, pp. 756-759, 2012