Font Size: a A A

Research On Diagnosis Technology Of Potato Disease Based On Leaf Image Analysis

Posted on:2020-11-21Degree:MasterType:Thesis
Country:ChinaCandidate:H YangFull Text:PDF
GTID:2393330599455409Subject:Engineering
Abstract/Summary:PDF Full Text Request
Potatoes are popular as a healthy and delicious food.In recent years,potato cultivation and consumption have been greatly increased nationwide.China has developed to the largest country in potato cultivation and manufacturing industry,but the development of large-scale potato cultivation industry is facing new difficulties and challenges.In the process of potato growth,due to the influence of soil,temperature and air in the planting area,it is easy to produce pests and diseases in the process of potato growth,which hinders the development of potato industry.How to diagnose and treat potato diseases and insect pests simply and quickly is an urgent problem.Image recognition technology is an important technology that emerges and develops in the environment of rapid development of information and continuous improvement of intelligence level.The main principle of image recognition is to find and process a large amount of information needed by production by computer instead of manual labor,so as to improve labor productivity and liberate workers from alternative work.Now the production is intelligent.Potato pest detection based on image recognition technology is a process of using image recognition technology to detect the occurrence of pests and diseases in the growth process of potatoes.The application of image recognition technology in the diagnosis of potato diseases can simplify the identification process and improve the recognition accuracy,thus promoting the increase of potato production.In this paper,the application of image recognition technology to the detection of potato pests and diseases is divided into three steps:(1)using OpenCV technology to gray,binary,denoising image preprocessing;(2)using sliding windows to extract the set feature values on the image,each sliding window to extract 20 feature values.The genetic programming algorithm is used to select the feature set and get the optimal feature subset;(3)KNN,SVM,random forest and logistic regression classification algorithm are used to identify the pests and diseases of potato,and the recognition results are obtained.In order to verify the effectiveness of the proposed method,a total of 72 samples of diseases and insect pests were presented,and three different diseases were divided into two categories,disease and disease-free,and two groups of comparative experiments were carried out.The first group of experiments used KNN,SVM,random forest,logical regression and other classification algorithms to experiment on feature sets without feature selection.Among them,the accuracy of the SVM algorithm was 90.52%,and the accuracy of KNN was 88.70%.In the second group,KNN was used to classify the feature sets that had been selected,and the accuracy was 90.20%.Through experiments,we can see that each classification algorithm has a better effect on disease diagnosis based on potato leaf image.After feature selection,the number of features can be reduced and the accuracy of classification and recognition can be improved.Using image recognition,the disease diagnosis of potato can be realized simply and quickly,and the prevention and treatment scheme of potato diseases and insect pests can be formulated,so as to increase the growth and income of potato and promote the rapid and healthy development of potato industry.
Keywords/Search Tags:pest detection, image recognition, feature extraction, feature selection, genetic programming
PDF Full Text Request
Related items