Remote Sensing Monitoring Of Winter Wheat Powdery Mildew Based On Hyperspectral And Multispectral Data

To quickly and accurately monitor the occurrence and development of wheat powdery mildew in large areas,in this article,three kinds of experiments were carried out using hyperspectral and multispectral data respectively..In the first part,in order to fully understand the spectral response characteristics of winter wheat powdery mildew,hyperspectral data was used to analyze the powdery mildew of winter wheat.Sensitive feature were selected by combining Relief-F algorithm with the correlation coefficient around the band from 400nm to 800nm.The selected features were 636nm and 784nm.A linear regression model,support vector machine(SVM)model and least-squares support vector machines(LS-SVM)model were used to classify winter wheat powdery mildew in this article.The experiments were carried out respectively from the classification of disease samples and the classification of all samples.The classification results of disease grade of disease samples were as follows:the one-dimensional regression model was suitable for classification of disease grade,while the SVM model and LS-SVM model were suitable for all samples classification.SVM classification accuracy is generally higher than LS-SVM.From the point of view of program running time LS-SVM run more efficiently,the larger the running sample,the greater the time difference between the two models._In other words,a one-dimensional linear regression model is more suitable for uneven distribution of samples.Generally speaking,LS-SVM is more suitable for actual production needs.Then second,wheat powdery mildew is one of the main serious diseases for winter wheat.A fast and accurate monitoring of the disease at a regional scale plays a vital role in reducing yield loss.Remote sensing data has great advantages over traditional data in disease monitoring,such as simpler operation,more real-time and higher resolution.In this study,Chinese HJ-1A/1B data with high revisit frequency and 30m spatial resolution was used to inverse Land Surface Temperature(LST),extract four-band reflectance data and build seven vegetation indices.These indices should be filtrated to improve accuracy of the model due to redundancy of them.Then,we implemented screening features with the combination of Relief and K-mean algorithm.Relief algorithm which can provide the basis for feature evaluation,so features were ranked in descending order judged by feature weights in preparation for the next process.Clustering accuracy obtained by K-mean algorithm.According to the weight of the feature,the features clustered in turn to perform K-mean analysis.Then the cluster with the highest precision was picked out,and we finally got the normalized difference vegetation index(NDVI),Simple vegetation index(SR)and surface temperature(LST)as the feature set.Wavelet feature can decompose the data in multi-scale and multi-direction,which can highlight the sensitive factor of vegetation index to a certain extent.40 wavelet functions were constructed from 5 scales and 8 directions,and made them convolve with features.Because there were too many wavelet features after convolved,the independent T-test samples were used to obtain the most sensitive wavelet feature of disease and the corresponding wavelet kernel function.After this process,three features corresponding to vegetation indices were available.These three wavelet features were used as input variables of the model.Support vector machine is a kind of machine learning method based on statistical learning theory.Its core idea is to minimize the structural risk by mapping the input linear indivisible data to the high dimensional space,which makes the difference between different samples.The class interval is the largest while the intra-class interval is the smallest,then the hyperplane is constructed to classify data.The monitoring model of wheat powdery mildew in Jinzhou City of Hebei Province was established by using support vector machine(S VM)with three groups of features.The first group used twelve vegetation indices as the input variables of the model,which served as a control group.The second one used three features after feature selection and the third used three features of the wavelet transform.Then the monitoring precision of the three models was compared and analyzed.The experimental results were as follows.The overall accuracy and the kappa coefficient of the third model(called GaborSVM)were 86.7%,0.583 respectively,have shown better performance over the first model(60%,0.286)and the second model(80%,0.444).These results showed that the combined method of wavelet analysis with SVM(GaborS VM)can be applied to large area disease monitoring based on satellite remote sensing image,and has important application value in improving the accuracy of disease monitoring.The third part is by using three kinds of commonly used satellites data to monitor and analyze of winter wheat powdery mildew.The three kinds of satellites are HJ satellite,Gaofen-1 satellite and Landsat8 satellite.The experimental results show tha NDVI obtained the highest classification accuracy in the twelve chosen vegetation indices,and the classification accuracy obtained by Landsat8 data was up to 93%,which is also the best combination in this experiment.The results show that different vegetation indices based on the same satellite data have different experimental results,and the same vegetation index based on different satellite data has different experimental results.Therefore,it is necessary to select the optimal combination according to the types of diseases to improve data utilization efficiency and test Accuracy.Physical and chemical characteristics of plants can be showed by spectral changes,it is therefore possible to extract wheat powdery mildew information from the spectrum.At present,the monitoring range of hyperspectral data has some limitations,then attempts to promote the characteristics of wheat powdery mildew in hyperspectral to multispectral data.The SVM model and Gabor are used to analyze the multispectral remote sensing data of environmental satellite,and the experimental results reach the requirements of mart.Due to the different satellite parameters,the constructed vegetation index shows different classification accuracy in the classification and identification of winter wheat powdery mildew.Therefore,we need to select the appropriate remote sensing data and the appropriate classification model according to different types of diseases in order to achieve the best test results.To increase wheat output and reduce economic losses.