Study On The Feature Extraction And Identification Diagnosis Of Strawberry Pests And Diseases Based On Image Processing

Crop is an essential resource for human production and life,and crop diseases and insect pests are the main factors affecting agricultural production.In order to solve the problem of untimely and inaccurate identification and diagnosis of strawberry pests and diseases,this paper took the common species of strawberry pests and diseases as the research object,and studied the identification and diagnosis of strawberry pests and diseases.The main work content is as follows:1.In order to overcome the interference of complex background,super green gray scale method and histogram equalization method are used for image enhancement,and bilateral filtering method is used for image denoising.2.Make full use of the RGB and HSV color models of images for spatial vector transformation,and use the k-means algorithm of index function to extract color features;In addition,different LBP algorithms are explored to extract texture features.Based on a large number of comparisons in the aspects of average accuracy,average precision,average recall,and time complexity,a combination of uniform LBP algorithm and rotation invariant LBP algorithm is proposed to extract texture features.3.Construct SVM classifiers with different kernel functions,select the best superparameters through ten-fold cross-validation,test on the test set,compare the average recall rate,average precision rate,average recall rate,confusion matrix,etc.,and improve the SVM classifier with good performance to improve the recognition accuracy.4.Designed the identification and diagnosis system of strawberry diseases and insect pests,and preliminarily realized the diagnostic function of strawberry diseases and insect pests.Results and conclusion: the pretreatment operation adopted in this paper has effectively realized the effect of removing noise and enhancing the image display.In order to can be more comprehensive disease spot image feature extracting,the integrated into the index function of k-means algorithm for extraction of color features,adopt the uniform rotation mode of LBP texture feature extraction algorithm,the results showed that the fusion algorithm for strawberry is normal,anthracnose,leaf spot,powdery mildew,starscream pests integrated average recall rate was 78.26%,the composite average precision is 81.06%,the average accuracy of 81.27%,the average elapsed time 21 s;Texture features and color features were spliced together to explore the accuracy of SVM.Linear combination of SVM classifier with weight of 0.32 polynomial and SVM classifier with weight of 0.68 gaussian kernel was used.The average recall rate,average recall rate and average accuracy were 96.306%,96.586% and 95.158%,respectively.