Efficient identification of Bachu mushroom by flourier transform infrared (FT-IR) spectroscopy coupled with PLS-GS-SVM

Abstract

In this study, a fast and effective classification method of Bachu mushroom is proposed based on Fourier transform infrared (FT-IR) spectroscopy and pattern recognition algorithms. After separately measuring the cap and stem of 35 Bachu mushroom samples, 47 coral fungi samples, and 39 Lentinus edodes samples, we set up two classification groups of Bachu mushroom-coral fungi and Bachu mushroom-Lentinus edodes, and then the obtained FT-IR spectral data of the mushroom powder was combined with PLS-GS-SVM and PLS-ELM modeling to test the performance of the two classifiers. The results showed that the classification accuracy of PLS-GS-SVM was 98.78 % for the cap data of the Bachu mushroom-coral fungi group and 98.78 % for the stem; touching the Bachu mushroom-Lentinus edodes group, the classification accuracy for the cap data is 98.78 % and 100 % for the stem. For PLS-ELM, the classification accuracy for the cap data was 94.87 % of Bachu mushroom-coral fungi group and 77.81 % for the stem; as for Bachu mushroom-Lentinus edodes group, the classification accuracy for the cap data was 99.45 % and 99.18 % for the stem. This study demonstrates FT-IR spectroscopy combined with statistical algorithms can accurately classify the samples of Bachu mushroom mixed with other mushroom. The study provides some basis for consumers and food inspection agencies to identify the authenticity of Bachu mushroom and for scholars to explore more universal sorting methods of mushroom.