Study On Detection Of Spores Of Rice Blast

In the grain production and consumption of our country,rice occupies an extremely important position and is one of the most important food crops in our country,so ensuring the safety of rice production is the key to ensure the food security of our country.Among the diseases affecting the safety of rice production,rice blast is a worldwide fungal disease,which is characterized by wide prevalence and great harm.Rice blast is a kind of gas-borne disease in plant diseases.Its transmission principle is mainly through the transmission of spores by air flow blowing.When the spores on rice plants reach a certain number,rice plants will be infected with rice blast.Therefore,the detection of the number of spores in the air can be used to predict and warn the occurrence of rice blast.At present,the traditional methods for capturing and identifying spores of rice blast disease mostly use spore traps to capture and sample,and bring the captured slides with pathogenic spores back to the laboratory for observation,and then use manual counting methods to carry out statistics.Due to the large number of samples and small spores,this method is timeconsuming and labor-intensive.And when the pathogenic spores in the sample are mixed with other pathogenic spores in addition to the rice blast spores,it will greatly increase the difficulty of observation and greatly affect the judgment of the initial disaster.To solve this problem,a portable spore capture instrument was developed and a rice blast spore image recognition model was constructed.The portable spore capture instrument can realize automatic,regular and multi-time field spore collection,and realize the capture of spores in different time periods of the monitoring plot;the rice blast spore image recognition model is based on image recognition technology to automatically recognize rice blast spore And statistics,has the advantages of fast,accurate and efficient.The specific research is as follows:(1)Portable rice blast spores capture instrument was developed as a field rice blast spores capture device.In the process of design and development,the practicability,portability,stability and simplicity of operation of the equipment are fully taken into account in order to realize the collection of spores in the field.The portable spore capture instrument includes a chassis with an air inlet and a capture air duct device,a single chip microcomputer control system.(2)Design of the portable spore capture instrument MCU control system.The single chip microcomputer control system is composed of driver,power conversion module,single chip microcomputer,battery,step motor and so on.The single chip microcomputer controls the step motor to drive the slide fixing plate to rotate,so that the slide on the slide fixing plate reaches the designated working position to start collecting rice blast spores;At the same time,the single chip microcomputer timing module starts timing,after reaching the set acquisition time,the single chip microcomputer controls the step motor rotation to make the next slide reach the designated working position to continue collecting spores.And so on,to be collected after the set number of slides stopped.(3)The process of image preprocessing includes image grayscale,image filtering and edge enhancement of image without sharpening mask.Image grayscale filters out the useless color information in the original image,which shortens the time of subsequent processing.Compares the noise reduction effect of gaussian filter and bilateral filter on spore image,and objectively evaluates the noise reduction effect of the two filters using peak signal-to-noise ratio(PSNR)theory.Edge enhancement processing of non-sharpening mask image highlights highfrequency information in the image,such as edge details,and improves the effectiveness of feature extraction.(4)Two different methods of spore feature extraction were studied.On the one hand,spore feature extraction based on Canny edge detection is studied,and the minimum outer rectangle aspect ratio,contour area,circumference,ellipticity and other morphological feature parameters of spores are obtained by edge detection,contour extraction and feature extraction.On the other hand,the method of spore feature extraction based on HOG feature is studied,and the feature is formed by calculating and statistics the gradient direction histogram of local area of image.(5)Kernel function is one of the key parameters to determine the detection efficiency of support vector machine model.In order to ensure the accuracy and efficiency of the constructed support vector machine model,the kernel function of support vector machine model is compared.The SVM model based on linear kernel function,radial basis kernel function and additive cross kernel function is constructed respectively.(6)The portable spore capture instrument in this study can realize the automatic,timing and multi-time capture of rice blast spores,and carry very convenient,can realize the effective capture of spores in scattered,numerous and geographically remote plots.the SVM model constructed in this study can realize the effective recognition and counting of rice blast spores,and the accuracy of recognition is about 98%,which meets the requirements of accurate and efficient detection of rice blast spores.