Comparison of short-wavelength infrared (SWIR) hyperspectral imaging system with an FT-NIR spectrophotometer for predicting alpha-amylase activities in individual Canadian Western Red Spring (CWRS) wheat kernels

Abstract

Alpha-amylase activity in individual Canadian Western Red Spring (CWRS) wheat kernels was predicted using spectral information across the wavelength region 1235–2450 nm. Reflectance spectra were collected from an SWIR (short-wavelength infrared) hyperspectral imaging system and absorbance spectra were recorded from an Fourier transform near-infrared (FT-NIR) spectrometer on the same kernels. The partial least squares (PLS) regression technique was used to model the alpha-amylase enzyme activity levels to the spectral information. The prediction accuracy varied with the pre-processing methods applied to the regressor and regressand. The highest coefficient of determination (r2) value obtained from the SWIR hyperspectral imaging system was 0.88 and 0.82 from the FT-NIR instrument. The imaging approach was more successful because it also had the advantage of being able to localise the region where spectra were extracted from.