Study On The Fast Detection By Spectral Image Of Protein Concentrations In Wheat Grains At Open Temperature Increasement

Protein concentrations is an important indicator to measure the nutritional and processing qualities of wheat grains,which are mainly affected by the factors including the temperature of wheat growth.In recent years,as global temperatures rising,the impact of warming on protein concentrations of wheat grains has become a research hotspot at domestic and abroad.Recent studies mainly used artificial climate box or open-top temperature-increasing equipment to simulate the warming environment,combined with conventional physical and chemical analysis to research the response of protein concentrations in wheat grains to environmental warming.The previous studies revealed the phenomenon of wheat grains protein concentrations to environmental warming,but it didn’t reveal the cause of it.In addition,conventional physical and chemical analysis methods are time-consuming and labor-intensive,which are difficult to meet the needs of agricultural production automation.In view of this,this paper used open temperature increasing equipment to simulate the temperature of different wheat growing environments and then applied near infrared and midinfrared spectra combined with chemometrics to predict the protein concentrations in wheat grains.Moreover,this paper used hyperspectral imaging to visualize the distribution of protein concentrations in wheat grains.The main research contents and conclusions were as follows:(1)The effect of open-air temperature increasing on wheat biomass and the protein concentrations of stems,leaves and grains during the wheat growthFirstly,infrared warming equipment was used to warm the whole process of wheat growth to simulate climate change(warming range is about 2 ° C).Combined with two farming methods,four treatment methods were proposed,which were labeled as conventional tillage with warming(CTW),conventional tillage without warming(CTN),no-tillage with warming(NTW)and no-tillage without warming(NTN).Secondly,the leaves,stems and grains were collected at the main growth stage of wheat,and the protein concentrations were determined.And in each one of five stages,the protein concentrations in wheat leaves,stems and grains of each batch were measured.Finally,the leaf area index and the length of growth period in the main growth stage of wheat were determined.The results indicate that warming accelerates the rate of nitrogen transfer from wheat stems and leaves to the grains and the protein synthesis rate of the grains during the filling period to the mature period.During warming treatment,the wheat leaf area index and growth period length also raises,which indirectly improves the protein synthesis rate in the grains.(2)Qualitative discrimination and protein concentrations determination of wheat grains with different treatmentsFirstly,wheat grains with different treatments were collected,and the near-infrared and mid-infrared spectra of the grains were collected respectively,and used different spectral pretreatment methods(SNV,MSC,SG,1stDER,2ndDER)to pretreat the collected spectra,and selected the best one.Secondly,the Kjeldahl method was used to determining the wheat grains protein concentrations.Finally,principal component analysis(PCA)was performed on the collected spectra.The results indicate that the spectra after SNV pretreatment can effectively reduce noise and improve the accuracy of the model.Under normal tillage and no-tillage conditions,warming increases the protein concentrations in wheat grains by 9.92% and 12.7%(P <0.05).After analyzing the spectrum with PCA,the near-infrared spectra can more effectively distinguish the batches of wheat with different treatments,compared with mid-infrared spectra.(3)Rapid determination of wheat grains protein concentrations based on spectral technologyFirstly,five different protein concentrations prediction models include iPLS、BiPLS、SiPLS、BP-ANN、LS-SVM were established based on near-infrared and mid-infrared spectra.Secondly,the data fusion method was used to fuse the selected characteristic spectra of near infrared and mid infrared spectra.Finally,the above models were established based on the fused spectra.The results indicate that LS-SVM is the best one of different prediction models under the same spectral technique.Specifically,the LS-SVM prediction model based on near infrared spectra has Rc of 0.9652,Rp of 0.8516,RMSECV of 0.214,and RMSEP of 0.448.The LS-SVM prediction model based on mid-infrared spectra has Rc of 0.9642,Rp of 0.8017,RMSECV of 0.235,and RMSEP of 0.468.The LS-SVM prediction model based on fusion spectra has an Rc of 0.9723,Rp of 0.9013,RMSECV of 0.197,and RMSEP of 0.348.(4)Based on hyperspectral imaging to visualize the distribution of protein concentrations in wheat grainsThe research object is a single wheat grain.Firstly,the spectral signals and physicochemical values of wheat grains with different treatments were extracted.At the same time,different spectral pretreatment methods(SNV,MSC,SG,1stDER,2ndDER)were used to pretreat spectra,and the best pretreatment method was selected.Secondly,five different protein concentrations prediction models include iPLS、BiPLS、SiPLS、BP-ANN、LS-SVM were established.Inputing the spectra of each pixel of the test single wheat grain into the optimal model to obtain the distribution map of wheat grain protein concentrations.Finally,the two-dimensional pseudo-color image was reconstructed in three dimensions to obtain the three-dimensional pseudo-color image.The results indicates that SNV is the best spectral pretreatment method.LS-SVM is the best protein quantitative prediction model,in which Rc is 0.9978,Rc is 0.9361,Rp is 0.9013,RMSECV is 0.315,and RMSEP is 0.327.The two-dimensional pseudo-color graph can be used to visually observe the protein concentrations of a single wheat grain,and the three-dimensional pseudo-color graph can be used to visually observe the difference in protein concentrations of a single wheat grain under different treatments.This article revealed the reasons why the increase in the temperature of the wheat growth environment leads to the increase in grain protein concentrations.At the same time,based on the spectra technology,the protein concentrations of wheat grains can be quickly analyzed.Moreover,combined with the hyperspectral imaging,the distribution of protein concentrations in a single wheat grain were visualized.The research conclusion can provide a theoretical basis for wheat production and testing.