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.

Thickness is one of the important appearance qualities of vegetable soybeans in the commodity classification. But conventional classification methods for vegetable soybeans are time-consuming and may not be optimal. As a nondestructive and rapid method, hyperspectral imaging technology has been investigated for classification of vegetable soybeans. The research evaluated the classification for vegetable soybeans which is determined by the thickness measurements using the hyperspectral imaing technology. Hyperspectral reflectance images of two hundred vegetable soybean samples between 400 and 1,000nm were acquired using a hyperspectral reflectance imaging system. A fuzzy-rough set model based on the entropy algorithm is proposed to select the optimal wavelengths for the hyperspectral imaging data of vegetable soybeans. Moreover, partial least squares discrimination analysis (PLSDA) model was developed for two-class ('thinness' and 'thickness'). The validation models yielded 85% and 97% classification accuracies for 'thinness' and 'thickness' with full wavelengths (400-1,000 nm), respectively. Better results, 94% and 97% for thinness and thickness, were obtained by validation models with three optimal wavelengths using the fuzzy-rough set algorithm. The simulation results demonstrate that hyperspectral imaging technology is potentially useful for classifying of vegetable soybeans.