Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Abstract. In order to keep the variety purity of corn seeds, hyperspectral reflectance imaging and chemometric methods were combined to classify three different corn seed varieties (Chunhua201, Jiayu538, and Qianfeng258). In this paper, 5 points and 3 times smoothing (5-3 smoothing) was used to preprocessing the spectra data, successive projection algorithm (SPA) was used to select the feature wavelength. The modeling performance of two chemometric methods (K nearest neighbors (KNN), support vector machine (SVM)) were compared. It was concluded that SVM model performed better than KNN model, the optimal results is 5-3 smoothing-SPA algorithm, the 9 feature wavelengths, which can get a 91.6% accuracy rate of the calibration set, and a 96.8% accuracy rate of the validation set.