Nargesi Mohammad Hossein, Kheiralipour Kamran
Mechanical Engineering of Biosystems Department, Ilam University, Ilam, Iran.
Heliyon. 2024 Aug 8;10(16):e35944. doi: 10.1016/j.heliyon.2024.e35944. eCollection 2024 Aug 30.
Adulteration detection in plant-based medicinal powders is necessary to provide high quality products due to the economic and health importance of them. According to advantages of imaging technology as non-destructive tool with low cost and time, the present research aims to evaluate the ability of the visible imaging combined with machine learning for distinguish original products and the adulterated samples with different levels of chickpea flour. The original products were black pepper, red pepper, and cinnamon, the adulterant was chick pea, and the adulteration levels were 0, 5, 15, 30, and 50 %. The results showed that the accuracies of the classifier based on the artificial neural networks method for classification of black pepper, red pepper, and cinnamon were 97.8, 98.9, and 95.6 %, respectively. The results for support vector machine with one-to-one strategy were 93.33, 97.78 and 92.22 %, respectively. Visible imaging combined with machine learning are reliable technologies to detect adulteration in plant-based medicinal powders so that can be applied to develop industrial systems and improving performance and reducing operation costs.
由于植物性药用粉末在经济和健康方面的重要性，对其进行掺假检测以提供高质量产品是必要的。鉴于成像技术作为一种低成本、省时的无损检测工具的优势，本研究旨在评估可见光成像结合机器学习区分原装产品和不同鹰嘴豆粉掺假水平的掺假样品的能力。原装产品为黑胡椒、红辣椒和肉桂，掺假物为鹰嘴豆，掺假水平分别为0%、5%、15%、30%和50%。结果表明，基于人工神经网络方法对黑胡椒、红辣椒和肉桂进行分类的分类器准确率分别为97.8%、98.9%和95.6%。采用一对一策略的支持向量机结果分别为93.33%、97.78%和92.22%。可见光成像结合机器学习是检测植物性药用粉末掺假的可靠技术，可应用于开发工业系统，提高性能并降低运营成本。
