Canopy Image Spectral Monitoring And Fruit Quality Analysis Of Hami Melon In Xinjiang By Drone Remote Sensing And Visible Near-infrared Spectrum

In this study,a field cantaloupe in Santanghu town,Hami,Xinjiang,was used to obtain canopy images of cantaloupe crops,at the same time,the relative chlorophyll content(SPAD)of canopy leaves was measured by hand-held chlorophyll meter.The spectral data of cantaloupe and cantaloupe were obtained by spectral technology,and the moisture content and soluble solids in cantaloupe were measured by drying and destructive sampling.Then,after feature extraction,the linear and nonlinear models are used for quantitative prediction modeling,and the optimal prediction model is determined by comparing and analyzing the results.The results show that the application of remote sensing and spectrum technology can provide a new method for the fast development of field crop physical and chemical parameters detection.The specific methodology and conclusions of the study are as follows:(1)For the pre-processed visible light images,15 kinds of visible light vegetation indices and 6 kinds of texture feature parameters were extracted,and then Principal Component Analysis(PCA)was used to reduce the dimension of the data and eliminate the variables with poor correlation,then,combining four kinds of linear and nonlinear models,the chlorophyll inversion model is established.The results show that when the first 5 principal components of the Dataset are extracted by Principal Component Analysis(PCA),the PCA-SVM model has the best prediction effect and the Correlation Coefficient of the prediction set is0.915.(2)Firstly,the fresh leaves collected from the cantaloupe canopy were collected in time,and the moisture content of cantaloupe leaves was measured by drying sampling method.Then,the feature wavelengths which are highly correlated with leaf moisture content in full band spectral data are extracted by using the combined sub-interval least squares and the combined algorithms of CARS and SPA.GA and PSO are used to optimize the connection weight(W)and the hidden layer neuron threshold(B)between the input layer and the hidden layer in the ELM model,where the optimal kernel function and the number of hidden layer neurons have been determined,the optimal and stable W and B values are obtained to further improve the stability and prediction accuracy of the model.Finally,four feature wavelength extraction algorithms are combined with ELM,GA-ELM and PSO-ELM to analyze the model,and the Correlation Coefficient between the correction set and the prediction set is taken as the evaluation index of the model.Finally,it is found that the prediction precision of leaf moisture content in cantaloupe canopy using GA optimized ELM model combined with Si PLS-CARS is the best,so Si PLS-CARS-GA-ELM is the best model for leaf moisture content retrieval,the R_C value is 0.9289 and the R_P value is 0.9032.(3)The spectral data of Hami Melon were obtained by spectral technology,and the soluble solids in Hami melon were detected by saccharimeter,then the spectral data of Hami Melon were selected by feature interval and extracted by feature variable,respectively,then combining(ELM)with(PLS),the prediction model of Soluble solids content in Hami melon was established.The results showed that the prediction model of Bi PLS+SPA+PLS was the best.The Correlation Coefficient of the corrected set and the predicted set was0.9234 and 0.8788 respectively.The model could accurately predict the content of soluble solids in Hami Melon.The above results provide theoretical support for monitoring the cantaloupe canopy in Xinjiang field.So as to guide fertilization for the actual nutritional status of cantaloupe,it can increase the efficiency and precision of fertilization management,promoting the development of precision agriculture,and provide a theoretical research method to improve and guarantee the quality of Hami Melon.