Visual detection of moldy peanut kernels based on the combination of hyperspectral imaging technology and chemometrics

Authors:
Publisher website
Google Scholar
Abstract
In order to facilitate the identification of moldy peanuts using hyperspectral imaging, in this study, hyperspectral images of four peanut kernels in a large dark box were acquired at the band range of 1,000–2,030 nm for classification of moldy peanut kernels. Then, the spectral data were extracted from hyperspectral images. Moreover, successive projection algorithm (SPA) was used to select effective wavelength. Next, three models including partial least squares discriminant analysis (PLS‐DA), support vector machine, and linear discriminant analysis (LDA) were established to identify moldy peanut kernels based on the optimal wavelengths selected by SPA. As a result, SPA–LDA generated the best effect with accuracy of 100% in both calibration set and prediction set. Finally, based on the SPA–LDA model, we successfully completed the visual detection of peanut seed mold using 1,120 nm wave band and threshold method. The results indicated that the combination of chemometrics and hyperspectral imaging technology provided an accurate, rapid, and nondestructive detecting method for the classification of moldy peanut kernels.Practical Applications: It is well‐known that the moldy peanuts contain potent carcinogen. Therefore, the accuracy of moldy peanuts identification becomes very important. Traditional methods for detection of moldy peanuts are tedious, time‐consuming and strict requirements for operators, which cannot meet the requirements of intelligent agriculture. Hyperspectral imaging technology, providing both spectral and spatial information simultaneously, can detect moldy peanuts rapidly, accurately, and nondestructively. All these results show that the use of hyperspectral technology combined with appropriate chemometrics can facilitate the accurate identification of moldy peanuts, and the high accuracy of discrimination also provides theoretical support for the rapid and accurate detection of industrial application of moldy peanuts in the future.
Funding Information
  • National Natural Science Foundation of China (31772068)

This publication has 30 references indexed in Scilit: