Identification Of Boletes Species And Origin Based On Near-infrared Two-dimensional Correlation Spectroscopy

Wild edible boletes have a delicious taste and unique flavor,and its fruiting body has the nutritional characteristics of high protein and low lipid,rich in polysaccharides,polyphenols,flavonoids,dietary fiber,vitamins and other active substances,with antioxidant,auxiliary hypoglycemia,regulate intestinal flora,enhance immunity and other health care functions,with high food and medicinal value.Its food and medicinal value is easily affected by factors such as species and origin,and the market price also depends on these factors to a large extent.At present,several links in the supply chain of boletes are often used to make illegal profits,which seriously infringes on the interests of consumers and disrupts the edible mushroom market order.The rapid identification of different species and origins of boletes is the focus of attention of edible mushroom market supervision.In this study,to address the problem of species and origin identification in the development of boletes industry,the near infrared(NIR)spectral data of common boletes from Yunnan were analyzed by chemometrics to identify different species and different origins of boletes,in order to: establish a rapid nondestructive identification model of boletes through NIR spectroscopy,so as to maintain market stability and development,protect people’s health and life safety,and provide a theoretical basis for the further development and utilization of green analytical techniques.(1)This study proposes a species identification method and optimizes it to maintain market order and protect the economic benefits of wild edible mushrooms.Two-dimensional correlation spectroscopy(2D-COS)images were generated based on the NIR spectra of boletes,which were used to build a deep learning model and optimize the recognition effect of the model.A total of 1398 spectral data of six species of boletes were collected,and the recognition accuracy of deep learning models built with different image types,different modeling parameters and different data sources were compared.The results show that synchronous 2D-COS is the best image type for building a deep learning model,and the recognition effect is further improved when the learning rate is 0.01 and epochs are40,using the stipe and cap data.The method retains the complete information of the samples and can provide a fast and non-invasive method to identify boletes species for market supervision.(2)Lanmaoa asiatica,as one of the common boletes in the edible mushroom market and with high nutritional and economic values,is often exploited by unscrupulous traders to make improper profits.In this study,a rapid and accurate origin identification model of L.asiatica was established based on NIR spectroscopy,and important climatic variables were screened by competitive adaptive reweighted sampling(CARS)algorithm.In this study,497 samples of L.asiatica were collected from 20 township-level areas in Yunnan Province,China.First,spectral analysis and principal component analysis(PCA)were performed based on the NIR spectra of all samples.To understand the possible chemical composition of L.asiatica and to observe the classification trend of 20 types of township-level samples.Then,2D-COS images were generated using Matlab-R2017 a software,and residual convolutional neural network(Res Net)image recognition models were built.The accuracy of the training and test sets of the Res Net model was 100%,with a loss value of 0.052,indicating that the model is extremely accurate.In addition,feature variables were selected from 105 climate variables using the CARS algorithm.Four important variables(February,March and April precipitation and January minimum temperature)associated with the difference in NIR spectra of L.asiatica were obtained by the CARS algorithm.In conjunction with the geographic traceability results,how characteristic climatic factors affect NIR spectral information is discussed.The results of the study can provide a rapid and accurate method for identifying the origin of boletes for market surveillance,and provide innovative ideas for screening key climatic factors.