Robust Classification of Tea Based on Multi-Channel LED-Induced Fluorescence and a Convolutional Neural Network

Abstract

A multi-channel light emitting diode (LED)-induced fluorescence system combined with a convolutional neural network (CNN) analytical method was proposed to classify the varieties of tea leaves. The fluorescence system was developed employing seven LEDs with spectra ranging from ultra-violet (UV) to blue as excitation light sources. The LEDs were lit up sequentially to induce a respective fluorescence spectrum, and their ability to excite fluorescence from components in tea leaves were investigated. All the spectral data were merged together to form a two-dimensional matrix and processed by a CNN model, which is famous for its strong ability in pattern recognition. Principal component analysis combined with k-nearest-neighbor classification was also employed as a baseline for comparison. Six grades of green tea, two types of black tea and one kind of white tea were verified. The result proved a significant improvement in accuracy and showed that the proposed system and methodology provides a fast, compact and robust approach for tea classification.

Quantitative Detection of Extra Virgin Olive Oil Adulteration, as Opposed to Peanut and Soybean Oil, Employing LED-Induced Fluorescence Spectroscopy

“…However, the light sources used in [ 17 , 18 ] were lasers, which are narrow in bandwidths (typically less than 1 nm) compared to that of LEDs (typically ten to twenty nanometers). Another phenomenon that can be observed is that the longer the central wavelength, the wider the bandwidth [ 22 ]. Thus, though lasers of wavelength larger than 440 nm can be utilized for oil adulteration detection, LEDs of this band are not suitable, because their own spectrum may overlap with the fluorescence spectrum.…”

“…With the rapid development of semiconductor technology, LEDs, with more commercially available wavelengths, can now be offered with decent light intensity from 370 to 470 nm. This facilitates the replacement of the exciting light source of LIF, resulting in what is called LED-induced fluorescence spectroscopy, which is expanding its use in various types of food detection applications [ 16 , 21 , 22 ]. Compared with LIF, LED-induced fluorescence spectroscopy requires less maintenance skills, smaller installation volume, and much lower cost, especially in the UV region.…”

Processing Fluorescence Spectra for Pollutants Detection Systems in Inland Waters

“…Hydrocarbons are more or less long chains of carbon and hydrogens and the added elements depend on each manufacturer. This way the purpose of this decomposition is not to identify its constituent components or concentrations [25] but identify the spectrum showing by a measurement [15,18]. Performing a spectrum adjustment with Gaussians or Lorentzians [24,26] from a purely mathematical point of view, it is a very useful tool that can lead to valid identification.…”

“…The sample was irradiated by the light source so that the compound emitted a fluorescence spectrum that was collected with a system consisting of (1) an optical fiber without collimator, and (2) a mini-spectrometer connected to a PC running MATLAB®(MathWorks®, Natick, MA, USA) program, a common tool for this objective [10,[14][15][16][17], which was responsible for the processing and preliminary storage. The optical power of the spectrum was proportional to the amount of the substance present, and the evolution of HP-LEDs caused devices with more optical power to appear on the market, making calibration necessary.…”

An Online Tea Fixation State Monitoring Algorithm Based on Image Energy Attention Mechanism and Supervised Clustering (IEAMSC)

“…Currently, the major procedures used in the tea research includes: spectral analysis [9,10], electronic nose aroma analysis [11][12][13], chemical composition analysis [14,15], neural network analysis [16,17], and image-based texture analysis [18,19], etc.…”