پديدآورندگان :
Naderi Nima Department of Chemistry, Sharif University of Technology, Tehran, Iran , Dehbasteh Maryam Department of Chemistry, Sharif University of Technology, Tehran, Iran , Parastar Hadi h.parastar@sharif.edu Department of Chemistry, Sharif University of Technology, Tehran, Iran
كليدواژه :
Chemometrics , Rice , Smartphones , RGB images , PLS , DA
چكيده فارسي :
Due to the high global demand for rice as a staple food, especially in Asian countries, there has been a significant increase in rice adulteration. As a result, authenticating rice samples and detecting fraud has become a major concern in the food industry. To address this issue, it is crucial to utilize suitable instruments for data acquisition. In recent years, portable instruments have gained popularity over conventional analytical systems due to their reliability, feasibility, cost-effectiveness, and rapid response. Among handheld devices, smartphones are widely employed due to their RGB imaging capabilities, where color is decomposed and quantified into red, green, and blue components [1]. However, raw RGB images often have low resolution, negatively impacting signal throughput during image analysis. Therefore, chemometric tools are required for qualitative and quantitative analysis to improve signal quality, enhance signal-to-noise ratio (SNR), and handle imperfect and variable input data. In this study, 93 rice samples were provided from three provinces in Iran (Mazandaran, Gilan, and Golestan) and recorded their RGB images using a smartphone. After image processing (extracting the RGB spectrum of each image and reshaping the data), Partial Least Squares-Discriminant Analysis (PLS-DA) was used as a supervised classification method to analyze the obtained data. This algorithm works by maximizing the correlation (or covariance) between X matrices (instrumental response, i.e. spectra) and Y matrix, which includes continuous or absolute values [2]. The proposed approach successfully discriminated rice samples from three different provinces (Mazandaran, Gilan, and Golestan) based on their geographical origins, yielding promising results of high accuracy, sensitivity, and specificity (100%) for both calibration and cross-validation sets. This study demonstrated that smartphones, in combination with chemometrics, are a reliable method for discriminating rice samples, with significant implications in terms of food safety and quality control. This approach can provide fast, accurate, and cost-effective results, helping to ensure the authenticity and quality of rice products.
