Research On Fast Acqisition And Detection Of Crop Information Based On The Chlorophyll Fluorescence

Chlorophyll fluorescence technology is often used to detect the plant nutrients and life information as the probe of plant research in the facility agriculture.Based on the chlorophyll fluorescence technology,machine vision technology and digital image processing technology,the study project sets up a chlorophyll fluorescence image stimulate and acquisition system to capture chlorophyll fluorescence image information.The system can obtain pepper leaf chlorophyll fluorescence image(fluorescence image)and extract chlorophyll fluorescence image characteristic value(characteristic value)to establish a defection model of leaf nitrogen content based fluorescence image,which is fast,nondestructive and efficient.Analysis the relationship of pepper leaf among chlorophyll fluorescence image characteristic value,SPAD and chlorophyll fluorescence kinetics parameters(fluorescence parameters)and establish a detection model of SPAD based fluorescence image at the same time.The main research work of this paper are as follows:(1)The study project established a chlorophyll fluorescence image stimulate and acquisition system based on the crop actual growth environment,which set LED coaxial laser light source,CCD camera,filter,cargo platform,computer,slippery course,camera work platform as the main hardware.(2)The study project used MATLAB to conduct preproccessing,pseudo-color processing,and image characteristic extraction(25 totally)on the collected chlorophyll fluorescence images,inculding R,G,B,GRAY,H,S,V,R/B/G,G/R,R,R/B,B/G,G/B,R/(G+B),G/(R+B),B/(R+G),r,g,b,NID,ExR,ExG,ExG-ExR,Exr,Exg.(3)The study project established a detection model based on chlorophyll fluorescence images of pepper leaf nitrogen content.Analyzed the relationship of fluorescence image characteristic value and leaf nitrogen content by SPSS,extracted characteristic values' three principal components(PCA)to be the independent variables of partial least squares(PLS)model,BP neural network(BPNN)model,generalized regression neural network(GRNN)model,and multivariate linear regression(MLR),to establish the detection models of seedling stage,flowering stage and fruiting stage separately.The best nitrogen detection model is BPNN by comparing the performance of models.In the three stages,its correlation coefficient of model(Rc)was up to 0.945 and root mean square error(RMSEC)was below 0.091,its correlation coefficient of prediction(Rp)was over 0.0905 and root mean square error(RMSEP)was below 0.204,the prediction effect were ideal.(4)The study project establish hed a detection model of pepper SPAD value based on chlorophyll fluorescence images.Analyzed the difference of peppers SPAD values and chlorophyll fluorescence images characteristic values,extract characteristic values' three principal components as the independent variables by using PCA,established a detection model of leaves' SPAD in all stages.In these models,the RMSEC and RMSEP were below 0.2 and the models' error is also very small.The prediction correlation coefficients of seedling stage,flowering stage and fruiting stage are 0.883,0.901,0.907 respectively and the prediction effect was ideal.(5)The study project analyzed the relationship between chlorophyll fluorescence image characteristics value and chlorophyll fluorescence parameters of pepper leaves.It was established linear fitting and curve fitting models of chlorophyll fluorescence parameters and chlorophyll fluorescence image characteristic values respectively.In the curve fitting models,the values of fitting correlation coefficient(R)of F and Y(?)are over 0.94,the root mean square error of fitting(RMSE)are below0.01.The values of ETR is over 0.935,but RMSE is over 0.4.The values of Fm',F0,F0',Fm,Fv/Fm,qN and NPQ are small,and RMSE is big,which have big error.Results show that F and Y(?)had better fitting effect,and fitting fluorescence parameters of Y and F(?)can use fluorescence image characteristic values.The above research results realized the function of chlorophyll fluorescence image stimulate and acquisition,which can extract chlorophyll fluorescence images quickly.The study project also established fast detection model of nitrogen content and SPAD values of whole pepper growth process.By this way,it provides a new rapid method of nutrient information detection for the facilities agriculture and provides a new idea of chlorophyll fluorescence technology in agricultural application at the same time.