文摘
The feasibility of near-infrared (NIR) spectroscopy for rapidly determining crude starch in different Coix seed varieties (big Coix seed (BCS), small Coix seed (SCS), and translucent Coix seed (TCS)) was investigated. The physicochemical properties of the starches were compared. Partial least-squares regression (PLSR) was applied to establish quantitative models. Competitive adaptive reweighted sampling (CARS) and genetic algorithms were used as variable selection methods, and the results showed CARS-PLSR had the best performances. The BCS and SCS starches had higher water-binding capacity, swelling power, solubility, relative crystallinity, peak viscosity, breakdown, hardness, and springiness than those of TCS starch. All these differences resulted from the amylose content and interactions between the amylose and the other components. These results suggest that the NIR spectral technique has the potential to determine crude starch in Coix seed, and the properties of three Coix seed starches are differently ascribed to cultivar variation.