Detection of green pepper impurities based on hyperspectral imaging technology

Abstract

To date, the intelligent assessment of green pepper quality remains an open question, particularly in aspects of color, as impurities closely resemble green peppers. Here, the hyperspectral imaging technology was employed to acquire the original spectral and image information of green and impurities. Subsequently, the original information was processed, and then trained using the super vector machine (SVM), to construct the green pepper impurity detection model. After training, the constructed model achieved 100% accuracy in the training set and 89.7% accuracy in the testing set, which generally met the application requirements. Visualization images of the constructed model in the application of identification green pepper impurity were prepared and optimized, which significantly achieved relatively satisfactory outcomes. Findings of this case study revealed that the presented strategy would provide a theoretical basis for the intelligent processing of green pepper, especially accelerate the development of impurity detection technology.