Study On Early Detection Of Rice Sheath Blight Disease Based On Hyperspectral Imaging Technology

Rice is the main food crop in China.Realizing stable and high yield of rice is always the important task of grain production in China.However,rice yield is seriously affected by diseases every year,which seriously influences the goal of high and stable yield.Conventional rice diseases detection fails to detect rice diseases in real time,so it misses the best control stage of rice diseases,which increases the use of pesticides.In this case,a research method based on hyperspectral imaging technology for early detection of rice sheath blight disease is proposed in this paper.This paper applied hyperspectral imaging technology,and the early disease information of rice sheath blight was obtained rapidly and non-destructively.First,this paper used the hyperspectral images to obtain the spectral information and image information of the infected rice plants and the healthy rice plants.Based on spectral characteristics and image features,classification?modeling and recognition were performed.And then,in order to optimize the performance of the model,the chlorophyll content was proposed to be another feature of disease recognition,which was combined with spectral characteristics and image features to build models to compare the performance of each model.The paper for disease early prevention and increasing the yield of rice is of great significance,which can be a reference for precision agriculture.The main conclusions of this paper are as follows:(1)According to the spectral dimension of hyperspectral image,extracted the region of interest(ROI)of healthy and infected rice leaves,pretreated the spectral data of the region of interest,the pretreatments included SG smoothing?SG-1D?SG-2D?SNV and MSC,then established the linear discriminant analysis(LDA)and support vector machine(SVM)classification models.The result showed that the linear discriminant analysis(LDA)model with SG-2D pretreatment achieved the best performance,the correct recognition rate of the modeling set is 98.3%and the correct recognition rate of the prediction set is 95%.(2)After five kinds of pretreatments,extracted the feature wavelengths with the method of x-loading weights,then established the linear discriminant analysis(LDA)and support vector machine(SVM)classification models based on feature wavelengths.The result showed that the linear discriminant analysis(LDA)model with SG-2D pretreatment achieved the best performance,the feature wavelengths are 453 nm?512 nm?529 nm?535 nm?547 nm?558 nm?675 nm?686 nm?693 nm?698 nm?705 nm and745 nm.,the correct recognition rate is 97.8%in the modeling set and 95%in the prediction set.Moreover,the model performance based on x-loading weights is equivalent to that of the whole band.So,it can be used to identify the rice sheath blight disease with x-loading weights.(3)According to the image dimension of hyperspectral image,the principal component analysis,probabilistic filtering and second-order probabilistic filtering were proposed in this paper,then established the back propagation neural network(BPNN)and support vector machine(SVM)classification models.The result showed that the BPNN based on image principal component analysis achieved the best performance,the correct recognition rate is 90.6%in the modeling set and 93.3%in the prediction set.(4)According to the spectral and image dimension of hyperspectral image,the chlorophyll content was proposed to be another feature of disease recognition,which was combined with spectral characteristics and image features to build models to compare the performance of each model.Then established the back propagation neural network(BPNN)and linear discriminant analysis(LDA)classification models.The spectral features combining with chlorophyll content?image features combining with chlorophyll content and spectral?image features combining with chlorophyll content were proposed.The performance of spectral?image features combining with chlorophyll respectively both better than that used the spectral and image features alone.BPNN based on spectral features combining with chlorophyll content achieved the best performance,the correct recognition rate is 100%in the modeling set and 96.7%in the prediction set,also,this model achieved the best performance compared with all models in this paper.