Research And Implementation Of Tomato Disease And Insect Pest Detection Method Based On Improved SVM

As the most important grain crop in China,the yield of tomato can be seriously affected by dozens of conventional tomato diseases such as leaf mildew,late blight,black spot miner,powdery mildew and Mosaic.If the types of tomato diseases and insect pests can be correctly judged in early stage and targeted preventive measures can be made in time,the direct economic loss to tomato will be greatly reduced and crop yield will be increased.The key to the process is how to quickly and accurately detect tomato pests and diseases.The traditional detection methods of tomato diseases and insect pests are usually conducted by professional researchers with their own professional knowledge and experience.However,the traditional detection methods are quite limited in the number of technical personnel,detection rate,prevention area,detection accuracy and other aspects,which can not fully adapt to the requirements of China’s agricultural modernization and information construction.Based on this actual demand,this paper proposes an improved SVM based detection and recognition scheme for tomato diseases and insect pests,and designs and implements the tomato leaf disease and insect pest recognition software.Firstly,this paper introduces the research background of the subject,discusses the tomato leaf pests and diseases,data set acquisition and processing algorithms,including image segmentation,data enhancement,feature extraction method and SVM classifier selection.Secondly,the key algorithms of tomato leaf pest identification were studied and discussed.In this paper,a variety of identification models based on SVM support vector machine were used to test tomato diseases and pests,including the identification scheme of tomato diseases and pests based on CCL-SVM algorithm and VGG16-SVM algorithm.In the identification design of tomato diseases and pests using CCL-SVM method,the fusion data of CCL(CM,CCV,LBP)with color texture were used as the feature input values,and then RBF-SVM method was used to calculate the optimal parameters to obtain the CCL-SVM model and detect the tomato diseases and pests.In the detection method of tomato diseases and pests using VGG16-SVM method,the feature extraction part of VGG16 image data and the classifier were firstly separated,and then the SVM classifier was reconstructed through the analysis task,and combined with the image feature extraction part of VGG16,the VGG16-SVM model was obtained and the detection and recognition of tomato pests and diseases was carried out.The experimental results showed that compared with VGG16-SVM,the CCL-SVM proposed in this paper significantly improved the identification performance of tomato leaf pests and diseases,the recognition rate increased by 3.25 percentage points,and the test time was much lower than other models.The CCL-SVM model adopted in this paper,with short detection time,high recognition rate and low technical threshold,has proposed a new method for the rapid identification of tomato diseases and pests.Finally,the construction process of tomato disease detection and recognition software is discussed.The software adopts the combination of Python and Tensor Flow to realize the detection and recognition of tomato disease and insect pests.By uploading pictures of tomato leaf diseases and insect pests to match the data set,the classification of diseases and insect pests can be detected and identified,which greatly improves the requirements of convenience and practicality of identification.