Research And Application Of Longan Disease And Insect Pest Feature Extraction And Classification Based On Evolutionary Algorithm And SVM

With the continuous development of precision agriculture,machine vision technology has been more widely used in the field of agricultural production to increase the benefits of agricultural production.Machine vision has achieved significant results in the application of crop pests and diseases and agricultural product quality inspection,and has played an increasingly important role in precision agriculture.Based on the research results of pest detection and identification both at home and abroad,this paper improves the algorithmic process of pest and disease image detection and classification,and further optimizes the classification model.The classification and recognition model presented in this paper can classify and identify longan diseases and insect pests.The classification accuracy is high,and the efficiency is better than the traditional model.It has a positive effect on long-term plant protection and other agricultural production.The main research work of this paper is as follows:(1)Longan insect pest image data collection and image preprocessing.Shooting and collecting images of plant diseases and pests such as longan and litchi,and then conducting pest and disease classification experiments based on some pest and disease images.The research subjects included common longan diseases and insect pests such as Phytophthora capsici,anthrax,pupa,and litchi.In this dissertation,filtering is used to remove the noise from the disease images.In addition,a variety of cross-image enhancement algorithms are used to enhance the image processing,which makes the image meet the requirements of experimental classification and identification.(2)The queen mating evolution algorithm was introduced into image segmentation.Using the segmentation method of the mating algorithm of the queen bee mating evolution and the Otsu algorithm,the lesion region extraction experiment was performed on the longan disease image,and it was verified through experiments that the method presented in this paper has a significant effect on the image segmentation application of pests and diseases,and the segmentation time is less.(3)Extracting valid data information in the longan lesion area,in short,extracting and analyzing the color features,texture features,and shape features of the lesion,and then normalizing the obtained effective data and performing lesions on the lesions The type annotation provides effective characteristic data for the research of pest classification and identification.(4)Through the research of the classic SVM algorithm module,the queen mating evolution algorithm is integrated into the process of determining the SVM classification threshold,and the SVM classifier that can be used for longan pest identification is trained,and the trained longan disease classifier is applied to other longan diseases and pest pictures Perform classification recognition tests.The experimental results show that the proposed SVM classifier based on queen mating evolution algorithm not only has good robustness,but also has the advantage of high recognition rate and superior performance.This thesis is based on the theoretical research of the mating evolutionary algorithm of queen bee.Through in-depth study of computer image processing,the queen mating evolution algorithm is integrated into the process of image segmentation.At the same time,the queen mating evolution algorithm is integrated into the classification threshold of the SVM classifier,and finally passed through the longan.The feasibility of the method proposed in this paper was verified by the extraction of diseased spot and the classification and identification of longan pests.