Measurement of Single Soybean Seed Attributes by Near-Infrared Technologies. A Comparative Study

Abstract

Four near-infrared spectrophotometers, and their associated spectral collection methods, were tested and compared for measuring three soybean single-seed attributes: weight (g), protein (%), and oil (%). Using partial least-squares (PLS) and four preprocessing methods, the attribute that was significantly most easily predicted was seed weight (RPD > 3 on average) and protein the least. The performance of all instruments differed from each other. Performances for oil and protein predictions were correlated with the instrument sampling system, with the best predictions using spectra taken from more than one seed angle. This was facilitated by the seed spinning or tumbling during spectral collection as opposed to static sampling methods. From the preprocessing methods utilized, no single one gave the best overall performances but weight measurements were often more successful with raw spectra, whereas protein and oil predictions were often enhanced by SNV and SNV + detrending. ABSTRACT: Four near-infrared spectrophotometers, and their associated spectral collection methods, were tested and compared for measuring three soybean single-seed attributes: weight (g), protein (%), and oil (%). Using partial least-squares (PLS) and four preprocessing methods, the attribute that was significantly most easily predicted was seed weight (RPD > 3 on average) and protein the least. The performance of all instruments differed from each other. Performances for oil and protein predictions were correlated with the instrument sampling system, with the best predictions using spectra taken from more than one seed angle. This was facilitated by the seed spinning or tumbling during spectral collection as opposed to static sampling methods. From the preprocessing methods utilized, no single one gave the best overall performances but weight measurements were often more successful with raw spectra, whereas protein and oil predictions were often enhanced by SNV and SNV + detrending.

Protein content prediction in single wheat kernels using hyperspectral imaging

“… Agelet et al (2012) reported that working at shorter wavelengths, i.e. 900–1650 nm, did not give significantly different results in terms of prediction performance.…”

“…Previous literature dealing with NIR prediction of small grain weight reported slightly better performances for soybean kernel weight prediction, i.e. R 2 = 0.77–0.91 depending on the spectral pre-treatment, wavelength region selected and instrumentation used ( Agelet et al, 2012 ).…”

“…Previous work investigated the effectiveness of spectral pre-treatment for NIR and HSI. For example, Agelet, Armstrong, Clariana, and Hurburgh (2012) discussed that, whereas SNV has been shown to minimise scattering effects, the combination and interaction of several factors producing scattering cannot be successfully addressed by SNV, as their calibration quality did not improve with SNV pre-treatment for single seed NIR measurements on soybean. However, in our case SNV appeared to be one of the most effective spectral pre-treatments, which can be attributed to differences in the shape and surface properties of the wheat kernels regarding the light scattering during the analysis.…”

“…Previous research reported on the possibility of predicting kernel weight based on NIR spectroscopy measurements, but for different types of grains, particularly maize and soybean ( Agelet et al, 2012 , Baye et al, 2006 ), and little information has been found in the literature on wheat kernel weight prediction by NIR or HSI and therefore no direct comparison can be made One paper has been found for wheat, where single kernel weight was predicted by NIR, with a reported RMSE of 2.8–4.0 mg, using a NIR instrument designed for single seed characterisation so that the light scattering was minimised ( Armstrong, 2014 ).…”

Rapid and Non-Destructive Detection of Compression Damage of Yellow Peach Using an Electronic Nose and Chemometrics

“…The levels of compression damage detection results show that the best UVE-LS-SVM model had the highest RPD, at 2.139, indicating that the model is usable for screening [49]. Besides the detection of the damage level, the discrimination of the damaged fruit is also important for the industry.…”