Research on the Application of Molecular Image Processing in Rice Quality Inspection

Area of Science:

• Agricultural Science
• Spectroscopy
• Image Analysis

Background:

• Traditional rice quality assessment relies on subjective and inefficient human sensory judgment.
• Molecular imaging technology integrates spectral and image analysis for rapid, nondestructive, and accurate agricultural product evaluation.
• Consumer demand for high-quality rice necessitates advanced, objective detection methods.

Purpose of the Study:

• To develop a rapid, nondestructive method for rice quality assessment using molecular imaging.
• To identify rice varieties and quantify moisture and starch content in rice samples.
• To demonstrate the feasibility and accuracy of molecular imaging in rice quality evaluation.

Main Methods:

• Acquisition of molecular images from rice samples of four origins.
• Extraction of spectral, textural, and morphological features from rice regions of interest.
• Application of Principal Component Analysis (PCA) for feature selection and model development, including Support Vector Regression (SVR).

Main Results:

• Identification of nine key feature wavelengths using PCA, achieving 91.67% accuracy in initial assessments.
• Development of optimal models, including BCC-LS-SVR and PCA-SVR with RBF kernel, demonstrating high accuracy (R² up to 0.989) for variety identification and content quantification.
• Successful detection of starchy rice using molecular imaging, validating its potential for quality assessment.

Conclusions:

• Molecular imaging technology provides a feasible, rapid, and nondestructive approach for comprehensive rice quality assessment.
• The developed models accurately identify rice varieties and quantify critical components like moisture and starch content.
• This research offers significant potential for improving agricultural product quality control and consumer assurance.