Evaluating green tea quality based on multisensor data fusion combining hyperspectral imaging and olfactory...

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Food Science

Background:

• Instrumental evaluation of tea quality using digital sensors is gaining global attention as an alternative to human sensory panels.
• Individual sensors often lack the discriminant accuracy needed for comprehensive tea quality assessment due to limited information.
• Key factors in sensory tea evaluation were considered to integrate multisensor data.

Purpose of the Study:

• To develop a robust method for instrumental tea quality evaluation by integrating multisensor data.
• To compare the effectiveness of different supervised learning algorithms for multisensor data fusion.
• To enhance the accuracy of green tea quality grading using combined spectral, image, and olfaction information.

Main Methods:

• Hyperspectral imaging and spectroscopy were employed to capture spectral and image data across various bands.
• Principal Component Analysis (PCA) was utilized for dimensionality reduction of the spectral and image data.
• Supervised learning algorithms, including Linear Discriminant Analysis (LDA), K-Nearest Neighbour (KNN), and Support Vector Machine (SVM), were applied for classification and comparison.

Main Results:

• Spectral features in the near-infrared (NIR) region and image features in the visible-near infrared/near-infrared (Vis-NIR/NIR) region demonstrated high classification accuracy.
• Multisensor data fusion using a Support Vector Machine (SVM) significantly outperformed individual sensor data for evaluating green tea quality.
• The overall accuracy for the calibration set increased from 75% (single sensor) to 92% (multisensor), and for the prediction set, it rose from 78% to 92%.

Conclusions:

• Multisensory data fusion provides an accurate method for identifying six distinct grades of tea.
• The integration of spectral, image, and olfaction data significantly improves the reliability and accuracy of instrumental tea quality assessment.
• This approach offers a promising alternative to traditional human sensory evaluation for consistent tea grading.