In recent years,with the rapid development of our country’s economic,agriculture has been rapid development at the same time,people pay much attention to the output of crops.At present,how to improve the crop yields is become a major topic.The detection and identification of disease images has become an integral part of the disease image system,it’s play a vital role in crop disease diagnosis and prevention,and it can improve the crop yield and increase people’s income.Thus,the depth research of the classification and identification about the disease image will be have a very important significance.The correct identification of crop disease type is the premise of disease prevention and control.In this paper,three kinds of common image samples of apple leaves were collected by means of image acquisition device.Firstly,the image of the collected is preprocessed,and the histogram equalization and median filtering are used to realize the image enhancement and diagnosing.And then use the gray histogram of the disease image to determine the threshold by the bimodal method to finish the first rough segmentation.And then analyze the characteristics of the gray histogram of the disease image to fine-tune it,and make the corresponding improvement of the image segmentation algorithm.Thus,detection and segmentation of the disease image has been completed.Secondly,by analyzing the advantages and disadvantages of several color space,extract the color characteristics of the segmented disease image in the RGB and HSI color spaces.And then extract the texture features of the segmented disease image according to the gray level co-occurrence matrix.And the principal component analysis method was used to select the most representative characteristic parameters.The feature vectors are composed of these features which are simple and can reflect the essential characteristics of the disease.And build the sample feature database.Finally,the comparison of the two classification methods.A support vector machine model is proposed to more suitable for classification of disease images.And then design the classification algorithm,The particle swarm optimization algorithm is used to optimize the parameters of the support vector machine model.And compared the accuracy of recognition of disease images under different parameters.Select the parameter with high recognition rate to establish the recognition model.Experiments were performed with SPSS software and MATLAB programming.The classification recognition rate is 92.667%T by Fisher discriminant analysis,And then use the LIBSVM software package and the support vector machine model to classify the disease images.First,the extracted 29 eigenvalues of the disease images in this paper are not optimized,The classification rate of disease images is 89.3939%.Secondly,nine principal components were extracted by principal component analysis,The classification accuracy rate is 92.4246%.Finally,optimize the parameters of the support vector machine model,when c(28)1618.28,g(28)039866.0,the classification accuracy of the model is 96.969%.Achieved the expected results. |