Design Of Virtual Simulation Platform For Crop Disease Recognition System Based On Image Analysis

This study emphasizes mainly on the proposal of automatic crop disease recognition method,through the combination of the statistical features of plant leaf images and its meteorological data.The images of the crop leaves infected with various diseases were taken based on different meteorological factors,such as temperature,growth periods and humidity.Leaf image enhancement and spot image segmentation were carried out successfully through various processes of contour extraction,image morphology and region growing algorithm.Statistical features of color,texture and shape were extracted through image processing;the optimal meteorological features with the highest accuracy rate were obtained and selected by the attribute reduction algorithm,from each image of the infected crop leaf.The statistical features and the meteorological features of the image were combined to derive the fusion feature vector.The classification accuracy was evaluated through the adoption of probabilistic neural networks(PNNs).The results of the experiment were obtained from three diseased leaf image datasets of cucumber plant,namely;downy mildew,blight and anthracnose,and these showed that the crop diseases can be effectively recognized through the integrated application of leaf image processing technology,involving the disease meteorological data and PNNs classifier.The accuracy rate of recognition was above 90%,this shows that higher classification accuracy could be achieved from the use of PNNs classifier trained on the disease feature coefficients extracted from the crop disease leaves and meteorological data.This study also covers areas such as crop disease identification techniques and a brief study of cucumber plant and its leaf disease;hence its leaves were used for the study experiment.In this paper,some information about image processing for plant disease detection are analyzed,including the Advances in image processing for plant disease detection,Plant disease detection methodologies,Platform requirements analysis,Java,Spring Framework,Eclipse,Server Operation Platform and the web server software.Some information about The Cucumber plant diseases are analyzed,including the Propagation of cucumber and the cucumber leaf diseases.Cucumber leaf diseases include the alternaria leaf blight,anthracnose,aphids,aster yellows,bacterial leaf spot,cucumber green mottle mosaic,downy mildew etc.Some information about materials and methods for cucumber leaves image processing are analyzed,including image collection,cucumber leaves image processing,segmentation of image disease spot,extracting the statistical features of the crop diseases,aggregation of meteorological data,the extraction of feature parameters,normalizing the extracted features of the leafs,probabilistic neural networks(PNNs)classifier etc.Experimental results and analysis are given.The statistic feature vector of color,texture,shape and meteorological are applied in classifier.Based on the automatic crop disease recognition method,which combined the statistical features of leaf images and meteorological data,the experimental results are obtained in this chapter.The experimental results show that the proposed method is more effective and could be potentially used in real field recognition.In this paper,the roles of Java programming in web development on image processing techniques are given.This part include the Java Image processing techniques,Java Image processing techniques for estimating plant leaf area,Java technologies for web applications,Java web application development procedures and the development software.This proposed virtual platform has been used in Image Processing Lab in the Department of Information and Electrical Engineering at Shenyang Agricultural University as a part of undergraduate curriculum.