Intelligent Monitoring And System Development Of Diagnostic Index Of The Wheat Seedling Growth

Wheat(Triticum aestivum L.)is one of the main food crops in China,understanding the growth status of wheat in the seedling stage and carrying out schemes of high-yield cultivation are of great significance for increasing production.Diagnosis of the wheat growth usually through the method of expert consultation,technical staff guidance,which can guide the production more accurately,while the efficiency of these methods are low.Hence,under the background of rapid development of high technology and "Internet + Agriculture",it's particularly important to monitor the wheat growth and build a reasonable evaluation method by means of information.In this study,we built intelligent estimation models of diagnostic index of the wheat growth by combining the diagnostic indexes of wheat seedling stage(from the beginning of winter to the jointing stage)obtained in field and image feature parameters at the same time.These intelligent estimation models include LAI estimation model and SPAD value estimation model,tiller number estimation model,biomass estimation model and nitrogen content estimation model.Based on that,estimating system of diagnostic index of the wheat growth based on image feature parameters was developed.Firstly,we extracted image characteristic parameters which correlated well with LAI,SPAD value,tiller number,biomass,N accumulation based on computer image processing technology.According to the previous research progress and the results of this study,the parameters are summed up as R?G?B?NRI?NGI?NBI?GMR?GR?ExG?ExR?ExGMR?NDIg?NDIb?NDI?Hue and canopy cover(CC).Second,we found suitable image feature parameters for constructing the estimation model of diagnostic indexes of the wheat growth by analyzing the Pearson correlation between 16 image feature parameters and diagnostic indexes,and optimal estimation models were established by the method of multiple stepwise regression analysis.The results of the model construction are as follows:(1)LAI showed significant correlations with GMR during the overwintering stage,the estimation model fitted by multiple stepwise regression analysis was y=0.160×GMR-0.784 and its R2,RMSE were 0.765**,0.286,respectively.The relationship between G and LAI was extremely significant in the jointing stage and the optimal estimation model was y=-0.274xG+34.919,the R2 of the model was 0.483**,the RMSE was 1.420.Therefore,it is feasible to estimate LAI in wheat seedling stage by using image characteristic parameters.(2)SPAD showed significant correlations with NDIb during the overwintering stage,the estimation model fitted by multiple stepwise regression analysis was y=1.116×NDIb+39.932 and its R2,RMSE were 0.236*,1.320,respectively.The correlation between SPAD value and GR value of wheat in the jointing stage was the best and the correlation coefficient is-0.568.Finally,the estimation models were fitted by multiple stepwise regression analysis and it was y=1.16×NDIb+39.932.The R2 of the model was 0.454**,the RMSE was 2.410.Therefore,the image characteristic parameters can be used as an index to evaluate the SPAD value of wheat seedling stage.(3)The tiller number showed significant correlations with GMR during the overwintering stage,the estimation model fitted by multiple stepwise regression analysis was y=69.668×GMR-270.228 and its R2,RMSE were 0.793**,122.294×104·ha-1,respectively.G,Hue and tiller number was significantly correlated at the jointing stage,the estimation models was fitted by multiple stepwise regression analysis and it was y=10.158×Hue-31.661×G+2568.074.The R2 of the model was 0.589**,the RMSE was 187.266×104 ha-1.Hence,the image characteristic parameters can be used as an index to evaluate the tiller number of wheat seedling stage.(4)The correlation between GMR and wheat biomass was the best during the overwintering stage and the estimation model was y=72.641×GMR-341.259,the R2 of the model was 0.818**,the RMSE was 115.188 kg·ha-1.The biomass showed significant correlations with G at the jointing stage,the estimation model fitted by multiple stepwise regression analysis was y=-170.295×G+21506.329 and its R2,RMSE were 0.496**,873.773 kg·ha-1,respectively.The results show that the image characteristic parameters have significant effect on the estimation of wheat biomass.(5)The correlation between GMR and N accumulation was the best during the overwintering stage and the estimation model was y=3.433 xGMR-17.039,the R2 of the model was 0.782**,the RMSE was 6.005kg·ha-1.The N accumulation showed significant correlations with G,ExGMR,CC at the jointing stage.GMR and nitrogen accumulation reached a very significant correlation and the estimation model fitted by multiple stepwise regression analysis was y=16.162xNDI-13.483xExGMX-75.390 and its R2,RMSE were 0.638**,25.689 kg·ha-1,respectively.This showed that the image characteristic parameters had significant effect on the estimation of wheat biomass.Therefore,it is feasible to estimate N accumulation by using the characteristic parameters of the image in wheat seedling stage.The intelligent monitoring system of diagnostic index of the wheat growth was developed in this study under the support of the agricultural knowledge and estimation model.The idea of design is:clear structure,friendly interface,easy operation,smooth running and perfect function.The purpose is to provide convenient services for the general user and the system can make an intelligent,real-time,and accurate diagnosis of the wheat growth.