Garlic Origin Traceability and Identification Based on Fusion of Multi-Source Heterogeneous Spectral Information.

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Machine Learning

Background:

• Garlic's chemical composition, nutritional value, and sensory properties vary by geographic origin.
• These variations influence consumer preference and acceptance of different garlic varieties.
• Identifying garlic's origin is crucial for quality control and authenticity.

Purpose of the Study:

• To develop a robust method for identifying the geographical origin of purple-skinned garlic.
• To evaluate the effectiveness of spectroscopic techniques combined with chemometrics and machine learning.
• To establish a theoretical foundation for origin identification in agricultural products.

Main Methods:

• Collected purple-skinned garlic samples from five distinct regions in China.
• Analyzed garlic cell components using mid-infrared (MIR) and ultraviolet (UV) spectroscopy.
• Applied preprocessing methods (MSC, SG Smoothing, SNV), Genetic Algorithm (GA) for feature extraction, and machine learning algorithms (XGboost, SVC, RF, ANN).

Main Results:

• Individual spectroscopy models achieved high accuracy (UV: 99.73% with SNV-GA-ANN; MIR: 97.34% with SNV-GA-RF).
• Fusion of UV and MIR spectral data with chemometrics significantly improved origin identification accuracy.
• Models utilizing fused spectral data (SNV-GA-SVC, SNV-GA-RF, SNV-GA-ANN, SNV-GA-XGboost) achieved 100% accuracy on training and test sets.

Conclusions:

• The fusion of ultraviolet and mid-infrared spectroscopy data, coupled with chemometrics and machine learning, provides a highly accurate method for identifying garlic's geographical origin.
• This integrated approach offers a reliable strategy for tracing the origin of garlic and other agricultural products.
• The study demonstrates the power of combining spectroscopic data fusion and advanced analytical techniques for agricultural product authentication.