| Feasibility study on hyperspectral imaging system for the determination of rice amylose content |
| Paper ID : 1426-NICAME1402 |
| Authors: |
|
Seyed Mehdi Nassiri *1, محمد باقر یگانه2, داریوش زارع2, علی محمد شیرزادی فر2, محسن صالحی دیندارلو3, محبوبه فضایلی4 1Department of Biosystems Engineering Shiraz University 2بخش مهندسی بیوسیستم دانشگاه شیراز 3بخش علوم و مهندسی صنارع غذایی دانشگاه شیراز 4بخش علوم و مهندسی صنایع غذایی دانشگاه شیراز |
| Abstract: |
| Hyperspectral imaging system was presented as a non-destructive and rapid method for determining the amylose content of rice. The quality of rice grain is of great importance for producers and consumers, and amylose is one of the major factors affecting the aroma and taste of rice kernels. In this research, based on the combination of spectra of rice grains and chemometric methods, including principal component analysis and partial least square regression, a system for evaluating the amount of amylose has been developed. For this purpose, first, rice has been dried under different drying conditions to change the amount of amylose. Then, the spectra of dried rice samples were extracted from hyperspectral images. In order to improve the accuracy of prediction and reduce noises, a combination of different pre-processing methods and optimal spectra have been acquired. Results showed that increasing the drying temperature leads to increased amylose content. Additionally, this trend can be obtained from principal component analysis by choosing the pre-processing method of smoothing, second derivative and mean centering as well as choosing the optimal spectra (500-800 nm). From calibration data and cross-validation method, the partial least square regression model achieved the highest accuracy, at R2CV=91.95% and R2Cal=89.95%, when using the pre-processing method and optimal spectra. Moreover, it can be seen that the partial least square regression model with R2Cal=92.48% can correctly predict the amylose content of rice grains using the spectra. |
| Keywords: |
| Amylose, Hyper-spectral imaging, Rice |
| Status : Paper Accepted (Poster Presentation) |
