Multi-wavelength laser-induced fluorescence spectroscopy with KLDA and Fisher score feature selection for pollen...

Area of Science:

• Botany
• Spectroscopy
• Data Analysis

Background:

• Accurate pollen identification is crucial for allergy diagnosis and ecological studies.
• Traditional methods can be time-consuming and require specialized expertise.
• Laser-induced fluorescence (LIF) spectroscopy offers a rapid, non-destructive analytical technique.

Purpose of the Study:

• To develop and validate a method for classifying eight pollen species using multi-wavelength LIF.
• To evaluate the effectiveness of kernelized linear discriminant analysis (KLDA) for pollen spectral data.
• To determine the optimal excitation wavelength for pollen discrimination.

Main Methods:

• Acquired fluorescence spectra from eight pollen species using LIF with 266, 360, and 405 nm excitation wavelengths.
• Applied Fisher score for feature selection and KLDA for dimensionality reduction.
• Utilized a Random Forest algorithm for final pollen classification.

Main Results:

• Achieved 100% classification accuracy for all eight pollen species tested.
• The 405 nm excitation wavelength yielded the most distinct spectral features.
• Integrating data from all three wavelengths further improved classification performance.

Conclusions:

• Multi-wavelength LIF spectroscopy combined with KLDA provides a highly accurate method for pollen classification.
• The 405 nm wavelength is particularly effective for differentiating pollen types via LIF.
• This approach offers a robust tool for botanical identification based on spectral characteristics.