Study On Intelligent Diagnosis Key Technology Of Soybean Leaf Diseases Based On Image

In this study,using image processing technology and Intelligent information processing technology combined with plant protection expert knowledge,based on the Internet of Things technology,crops intelligent diagnosis and control technology,corrected and pretreated the nonlinear distortion images of diseased leaves of soybean.The disease region has been identified using genetic neural network,achieve the lesion characteristics extraction method.Complete soybean leaf diseases intelligent diagnosis quantum neural network based on agricultural the Internet of Things soybean disease remote diagnostics system builders Program.Specific activities include the followings:Firstly,Discusses the base on homemade portable data collection templates Daejeon complex background of nonlinear distortion of the image acquisition mode,study several problems in the pretreatment of plant images,at the same time put forward a new method to cut any enclosed area of soybean leaf images.Combine with phase consistency check and the area of ? ? mathematical morphology method,as a complex environment soybean diseases blade conversion for a simple background in order to facilitate the processing target image extraction method.Given the use of neighborhood average as geometry,color characteristic parameters extracted before the pretreatment and selection of median filter as a texture,spot analysis before pretreatment technology research.Secondly,to avoid video collection angle,Light and other effects,reduce image noise,enhance the useful disease feature information,For suboptimal conditions collected soybean leaf diseases geometric distortion of the image and color distortion correction techniques study.Study soybean leaf image distortion generation principle,put forward bilinear mapping equations to restore the original spatial relationship of the distorted image,proposed supervision color correction method based on the standard gray card,That is determine the soybean disease image color values to the standard light in non-standard light ransforming relationship of non-linear incremental,and gives the algorithm,Achieve suitable for crop Daejeon image,recovery new technologies fo istortion correction image.Thirdly,the basic theory of research soybean lesion divided by the region recognition,proposed genetic computing and feed forward neural networks Combine with adaptive genetic neural network model of nonlinear mapping method,as a pixel attributable to the decision-making reasoning system.And gives the error back propagation method method of gradient descent learning algorithm.This algorithm is an optimal genetic algorithm mainly by global search and combine with error back propagation local gradient descent optimality,the experiments show,this algorithm is berrer,Identifiable lesion regional images can well satisfy the need of disease diagnosis.Fourthly,based on digital image processing technology and soybean disease diagnosis knowledge of plant protection experts,study of the color image in the lesion region and feature extraction algorithm,carried out geometry,color and texture feature calculation,Completed the lesion size of the area,perimeter,complexity,spherical,the center of gravity,the length of the axial ratio,the average coefficient of variation,H mean,S mean,L mean,average gray,smoothness,third moment,consistency,entropy and fractal dimension etc,a total of 15 dimensional feature vector,provides the best basis of identification classification of diseases for the future.Fifthly,according to the factors of the soybean disease diagnosis,study in multiple quantum level if can activate Quantum Neural Network of the function Quantum Neural.And gives parameter learning algorithm,summary of the soybean leaf disease diagnosis model,designed soybean leaf diseases non-destructive intelligent diagnosis algorithm and process.This approach is achieed by the feature vector features in the feature space.Practice shows that achieved remarkable results in the actual diagnosis of soybean leaf diseases.Finally,research through the Internet of Things application of agricultural information analysis can know,build the overall structure of soybean diseases intelligent diagnosis of Internet of Things system,proposed video and portable data collection sensor node structure,gives the main control center functions,completed image acquisition system design program based on agricultural of Internet of Things.Finally identified soybean disease information database and disease knowledge building program.Use the tectonic theory and methods of the knowledge base and database,gives the Soybean diseases Remote Diagnosis Expert System technical measures and the overall design program.Based on the above research content and initial results,this study presents soybean leaf diseases remote intelligent diagnosis the original model and Results of implementation of the program environment of agriculture of Internet of Things.The study efficient accuracy,easy replicability,can meet production requirements,at the same tine the above study also provides a reference method for the classification and identification of other crop diseases.Appropriate for extensive agriculture research workers and producers better use and universal,Provides technical basis guarantee for the development of intelligent agriculture.