Loading...

Conventional and Hyperspectral Imaging Combined with Chemometric Techniques for Rice Authentication

Dehbasteh, Maryam | 2024

9 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 56854 (03)
  4. University: Sharif University of Technology
  5. Department: Chemistry
  6. Advisor(s): Parastar Shahri, Hadi
  7. Abstract:
  8. Fast, non-destructive and reliable methods for food authentication purposes have been deemed important in recent years. The hyperspectral imaging technique (HSI) is a powerful method in this field, allowing the study of different features such as size, color, texture. Comprehensive interpretation is crucial, and utilizing chemometric methods enhances data interpretation. Rice is a vital and strategic food source for nearly half of the world, with Asia, particularly Iran, being a leading producer and exporter. Various types of rice fraud exist, including the mixing of high-quality and low-quality varieties, sold at a high cost. The present project aims to determine the geographical origin and authenticate Iranian rice samples using HSI, comparing the results with RGB images. To achieve this, three approaches were considered. HSI data from 163 samples in the wavelength range of 400-1000 nm were obtained. The very first one was using the mean spectra of samples. Distribution of data was studied by (PCA) and then a supervised method, (PLS-DA), was used for classification. Supplementary to this, HSI data was binned and the same investigation was applied. Since the results of such approaches was not satisfying, the average accuracy of 84% and 63% respectively, the pure spectral and spatial profiles of selected components of rice samples were extracted using (MCR-ALS) algorithm. By creating the data matrix, its performance was determined by PLS-DA and DD-SIMCA algorithms which in this case PLS-DA results were promising. In the final step, RGB data was compared with HSI data. To do so, the HSI data was converted to RGB ones so that by remaining all variables constant, the difference between wavelength channels can be considered. For further investigation, the RGB data of 20 random samples were captured by a smartphone. PCA and PLS-DA were applied to this set of data too. The accuracy of both techniques fell between the range of 81% to 100%, demonstrating that imaging methods can be considered practical for food authentication
  9. Keywords:
  10. Hyperspectral Imaging ; Rice ; Partial Least Squares (PLS) ; Multivariate Curve Resolution ; Multivariate Curve Resolution-Alternative Least Square (MCR-ALS) ; Chemometrics Method ; Conventional Imaging

 Digital Object List

 Bookmark

No TOC