Cultivar Classification of Single Sweet Corn Seed Using Fourier Transform Near-Infrared Spectroscopy Combined with Discriminant Analysis

Abstract

Seed purity is a key indicator of crop seed quality. The conventional methods for cultivar identification are time-consuming, expensive, and destructive. Fourier transform near-infrared (FT-NIR) spectroscopy combined with discriminant analyses, was studied as a rapid and nondestructive technique to classify the cultivars of sweet corn seeds. Spectra with a range of 1000–2500 nm collected from 760 seeds of two cultivars were used for the discriminant analyses. Thereafter, 126 feature wavelengths were identified from 1557 wavelengths using a genetic algorithm (GA) to build simplified classification models. Four classification algorithms, namely K-nearest neighbor (KNN), soft independent method of class analogy (SIMCA), partial least-squares discriminant analysis (PLS-DA), and support vector machine discriminant analysis (SVM-DA) were tested on full-range wavelengths and feature wavelengths, respectively. With the full-range wavelengths, all four algorithms achieved a high classification accuracy range from 97.56% to 99.59%, and the SVM-DA worked better than other models. From the feature wavelengths, no significant decline in accuracies was observed in most of the models and a high accuracy of 99.19% was still obtained by the PLS-DA model. This study demonstrated that using the FT-NIR technique with discriminant analyses could be a feasible way to classify sweet corn seed cultivars and the proper classification model could be embedded in seed sorting machinery to select high-purity seeds.

Application of near-infrared hyperspectral imaging to identify a variety of silage maize seeds and common maize seeds

“…The important wavelengths between 1110 nm and 1380 nm might be attributed to the second overtone of C-H stretch. 38 The spectral band at 1405 nm might be attributed to the O-H stretch. 39 The spectral bands at 1460 nm and 1470 nm might be attribute to the rst overtone of N-H stretching.…”

Rapid Vitality Estimation and Prediction of Corn Seeds Based on Spectra and Images Using Deep Learning and Hyperspectral Imaging Techniques

“…For example, Fourier near-infrared spectroscopy (FT-NIR) and Raman methods are based on vibrational spectroscopy to take advantage of the differences in biochemical composition between aged and normal seeds [8,9]. These methods have also proved to be meaningful in the discrimination of other crop varieties [10] and the detection of internal components [11]. However, due to the characteristics of spot scanning, it is impossible to obtain all the information needed for full evaluations.…”

Application of Visible/Infrared Spectroscopy and Hyperspectral Imaging With Machine Learning Techniques for Identifying Food Varieties and Geographical Origins

“…In the food classification field, the commonly adopted supervised learning algorithms contain PLS-DA (68,69), SVM (43,70), LSSVM (53), RF (71), BPNN, RBFNN (19,52), extreme learning machine (ELM) (54) and newly introduced deep convolution neural network (DCNN) (72). Among these algorithms mentioned above, PLS-DA is one of the most widely implemented chemometric methods in VIS/IR spectroscopy analysis for the advantage of handling data with collinearity.…”

“…The categories of samples are dummy variables with only zero and one, and the cutoff usually be set as 0.5 (73). Besides, SVM is also a commonly used classification method, which can map the original data into higher dimensional spaces with kernel function, and it optimizes a hyperplane with an appropriate margin to classify different groups (43). Radial basis function (RBF) is a usually used kernel within SVM.…”

“…Therefore, the spectral ranges could be selected according to the practical condition. For example, the 833-2,500 nm, 908-1,672 nm, and 1,000-2,500 nm were adopted by Cui et al (38), Yu et al (42), and Qiu et al (43), respectively, and all achieved the accuracy of over 97% for maize varieties classification. There existed differences regarding spectra information collection between VIS/IR and HSI.…”