نوع مقاله : مقاله پژوهشی
نویسندگان
1 دانشجو دکتری، گروه مهندسی مکانیک بیوسیستم، دانشکده کشاورزی دانشگاه شهرکرد، شهرکرد، ایران
2 دانشیار گروه مهندسی مکانیک بیوسیستم، دانشکده کشاورزی، دانشگاه شهرکرد، شهرکرد، ایران
3 استادیار، گروه مهندسی مکانیک بیوسیستم، دانشکده کشاورزی، دانشگاه علوم کشاورزی و منابع طبیعی ساری، ساری، ایران .
4 دانشیار، گروه داروشناسی_سم‌شناسی، دانشکده داروسازی دانشگاه شهید بهشتی ، تهران تهران
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Rice is a vital and strategic product that is used as a major food source. The high demand for purchasing and consuming rice leads to the adulteration of this product globally. Hence, a non-destructive and rapid method is needed to verify the authenticity of rice. Hashemi rice, a high-priced and high-quality rice in the market, is combined with rice such as Neda and Shiroudi, which are very similar in shape but lower in quality and price than Hashemi rice. This study used hyperspectral imaging (HSI) coupled with a fuzzy clustering algorithm to assess adulteration in Hashemi rice samples. First, to reduce the data's dimensionality, the principal component analysis method was applied to the preprocessed data using the multiplicative dispersion correction and Savitzky-Golay methods. Then, the fuzzy unsupervised clustering algorithm was applied using the whole spectrum wavelength (400-1000 nm). It was able to separate the original sample from the adulterated samples well. Also, the fuzzy membership diagram separated the original and self-adulterated samples, mixing 5% to 50%, confirming the correctness and capability of the fuzzy method. Therefore, the HSI system with fuzzy unsupervised algorithms can be used as a reliable and out-of-laboratory method for rapid rice authenticity evaluation.
کلیدواژه‌ها [English]
EXTENDED ABSTRACT
After wheat, rice is the second strategic crop in the world and is one of the main sources of dietary fiber, fat, protein, and other rare nutrients. Starch, protein, and fat are the main components of rice grain. Due to the high rice demand, this product is very vulnerable to adulteration at the global level. Considering the increasing scale and nature of adulteration, using reliable and fast methods to deal with rice adulteration is necessary. Fast, non-destructive, and non-contact hyperspectral imaging (HSI) method is a combination of imaging and spectroscopic technologies. With this method, spatial and spectral information can be obtained simultaneously for each point of the rice sample. This study aims to develop the HSI method and the clustering method to investigate the ability of a fast and non-destructive method for authenticating and predicting the percentage of Iranian rice adulteration.
In this research, a model was developed to detect rice adulteration by HSI device along with the fuzzy clustering method. HSI is a non-destructive and fast method and in the wavelength range of 400 to 1000 nm, all original (Hashmi) and adulterated (Shiroudi and Neda) samples and combinations of fake rice with Hashemi rice from 5 to 50% were scanned. Then they were pre-processed using MSC and SG algorithms and the best wavelengths were selected and checked using evolutionary wavelength selection. Finally, using the FCM method, a model was presented to detect adulteration in rice.
The results showed that the clustering method (fuzzy) together with the HSI system (as a non-destructive, fast, and accurate system) for predicting adulteration in rice samples shows satisfactory results. PCA after applying MSC + SG pre-processing, three samples of Hashemi, Shiroudi, Neda and are well separated from each other and can be identified. PC1, PC2, and PC3 after MSC + SG pre-processing for Hashemi, Shiroudi, and Neda cultivars were calculated as 60, 17 and 7%, respectively, with a total explained variance of 84% which was distributed by three PC1, PC2, and PC3. Then, after PCA, the pre-processed spectra were used for the C-means fuzzy clustering model. In the fuzzy clustering results, the original sample (in green color) and the fake sample (in red color) can be seen, and the fuzzy membership chart was used to confirm the results and further analysis. Two Hashmi rice samples (original) in green color and the Neda sample and its counterfeits (counterfeit) in red color (100 pixels or the number of first samples) belong to original and counterfeit clusters with high membership degree (almost complete). After the amount of 100 samples (100-400), two samples of Hashemi and Neda are mixed, which proves the combination of these two samples (from 5% to 50%). Fuzzy results in the full wavelength of its transparency over the selected wavelength.
Results proved that the HSI system coupled with the fuzzy clustering method can satisfactorily evaluate the rice authenticity. The fuzzy membership graph can predict adulteration in rice samples well.
Mahsa Edris: Writing – original draft, Methodology, Data curation, Software, Formal analysis.
Mahdi Ghasemi Varnamkhasti: Review and edit, conceptualize, supervise, and manage project administration.
Sajad Kiani: Writing – review and editing, Formal analysis, Investigation, Data curation, Validation.
Hasan Yazdanpanah: Supervision, Project administration, Methodology, Resources.
Zahra Izadi: Supervise, project administration, methodology, and resources.
Data available on request from the authors. All the data used in this original research are presented throughout the text and in the form of Tables and Figures.
The authors avoided data fabrication, falsification, plagiarism, and misconduct.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
The authors extend their sincere appreciation to Shahrekord University, Sari University of Agricultural Sciences and Natural Resources, Rice Research Center of Iran, and Food and Drug Administration for their support throughout this project.
The author declares no conflict of interest.
)