An Automated Image Processing Module for Quality Evaluation of Milled Rice

Area of Science:

• Agricultural Technology
• Computer Science
• Data Science

Background:

• Rice quality assessment is crucial for the food industry.
• Traditional methods for rice quality evaluation can be time-consuming and subjective.
• Developing automated, low-cost systems is essential for widespread adoption.

Purpose of the Study:

• To develop a low-cost rice quality assessment system using image processing and machine learning (ML).
• To classify different rice varieties based on extracted structural and geometric features.
• To evaluate the performance of various ML algorithms for rice classification and pricing adulterated samples.

Main Methods:

• Acquisition of 3081 rice grain images using a Raspberry-Pi based module.
• Extraction of structural and geometric features (e.g., area, roundness, shape factor).
• Classification using seven ML algorithms, including Random Forest, with performance evaluation via ROC curves.

Main Results:

• The Random Forest classifier achieved the highest accuracy of 77% for rice variety classification.
• The developed system successfully segmented and classified rice grains based on their types.
• The Random Forest algorithm was also effective in determining the price of adulterated rice samples.

Conclusions:

• A low-cost, automated rice quality assessment system can be effectively implemented using image processing and ML.
• The Random Forest classifier is a robust model for classifying rice varieties and assessing sample value.
• This technology offers potential for improving efficiency and accuracy in the rice industry.