Research On System Of Rice Diseases And Pests Image Recognition And Diagnosis Based On Android

The first premise of the integrated management of rice diseases and pests is to identify and diagnose the diseases and insect pests in a timely and accurate way.At present,the identification and diagnosis of rice diseases and pests in our country mainly depend on the artificial identification of plant protection technicians,which is laborious and time-consuming and can not meet the development needs of modern agriculture.With the development of machine learning and pattern recognition technology,the automatic identification and diagnosis of rice pests and diseases become possible.In order to realize the automatic identification of rice pests and diseases in the field,this paper established an image recognition and diagnosis system of rice diseases and pests based on Android.The system can quickly identify and diagnose collected images of pests and diseases in the field.The main contents and results of this paper are as follows:(1)The rice pests and diseases information database is established by using and accessing the API interface of the SQLite database of the Android mobile phone,which contains the image information,the period of occurrence and the corresponding control measures of rice pests and diseases.It has guidance and popularization for farmers.(2)The Android-based image recognition and diagnosis system for rice diseases and pests is built,which includes client and server side.The client is an APP that can realize pest and disease image collection,information browsing and server-side communication;Server-side includes responding to the client request information and calling the corresponding recognition algorithm to classify the pest and disease image.It is a nondestructive,timely and accurate method of identification.Among them,the mobile phone with image cropping function,to reduce the amount of data transfer and shorten the system uptime is very helpful.The system of real-time,efficient,accurate and portable,can provide farmers with convenient and efficient identification and diagnosis of pests and diseases in the field.its application prospects are good.(3)The detection algorithm based on SVM for rice diseases and damage-like images was studied.Firstly,the image segmentation algorithm based on saliency is used to remove the background of the image.Then,the color,texture and shape features of the lesion images are extracted.Finally,SVM classifier is used to classify and recognize the four kinds of rice diseasesand damage-like images,with an average recognition accuracy of 92.0%.(4)The pest image recognition algorithm based on feature fusion and sparse representation is studied.Firstly,all pest images were preprocessed and their global color features and local features were extracted.The feature selection and fusion were carried to obtain the optimal features for pest classification.Next,a multi-feature overcomplete dictionary was constructed and each column vector represents a training sample.Then,the testing samples were sparsely represented by the overcomplete dictionary.Meanwhile,the sparse concentration index threshold value was used to determine the validity of the testing sample.If the sparse concentration index of a testing sample is greater than the threshold value,the pest was identified the species with minimizing reconstruction error.Otherwise,the testing sample was judged to be a non-forecasting pest.Finally,the sparse representation classifier based on the fusion of global color and HOG features could obtain better identification results.The identification rate was90.1%.The image recognition and diagnosis system of rice pests and diseases based on Android has real-time,high efficiency,accuracy and portability.It can provide farmers with convenient and fast method for rice field pests identification and diagnosis,and has good application prospects.