Related Experiment Video
Updated: Jan 4, 2026
06:56
Classification of Neural Stem Cell Activation State In Vitro using Autofluorescence
Published on: April 12, 2024
924
Robust Classification of Tea Based on Multi-Channel LED-Induced Fluorescence and a Convolutional Neural Network
Hongze Lin1, Zejian Li2,3, Huajin Lu4
1School of Automation, Hangzhou Dianzi University, Hangzhou 310018, China. linhongze@hdu.edu.cn.
Sensors (Basel, Switzerland)
|October 31, 2019
Summary
This study introduces a new method using a multi-channel light emitting diode (LED)-induced fluorescence system and a convolutional neural network (CNN) to accurately classify tea varieties. This approach offers a fast and robust way to identify different types of tea leaves.
Area of Science:
- Analytical Chemistry
- Spectroscopy
- Machine Learning
Background:
- Accurate classification of tea varieties is crucial for quality control and market differentiation.
- Traditional methods for tea classification can be time-consuming and subjective.
- Developing rapid, objective, and accurate classification techniques is an ongoing challenge.
Purpose of the Study:
- To develop and validate a novel system for classifying tea varieties using LED-induced fluorescence and a convolutional neural network (CNN).
- To compare the performance of the proposed CNN method against traditional Principal Component Analysis (PCA) and k-nearest neighbor (KNN) classification.
Main Methods:
- A multi-channel fluorescence system was designed using seven LEDs (UV to blue) as excitation sources.
- Fluorescence spectra from tea leaves were collected sequentially and merged into a 2D matrix.
- A convolutional neural network (CNN) model was employed for pattern recognition and classification of tea varieties.
- PCA combined with KNN was used as a comparative baseline method.
Main Results:
- The proposed LED-induced fluorescence system combined with CNN achieved high accuracy in classifying six grades of green tea, two types of black tea, and one type of white tea.
- The CNN method demonstrated a significant improvement in classification accuracy compared to the PCA-KNN baseline.
- The system proved effective in distinguishing between different tea varieties based on their fluorescence signatures.
Conclusions:
- The integrated system of multi-channel LED-induced fluorescence and CNN provides a fast, compact, and robust approach for tea classification.
- This methodology offers a promising alternative to conventional tea analysis, enhancing objectivity and efficiency.
- The study highlights the potential of machine learning in conjunction with spectroscopic techniques for agricultural product authentication.
