Kiwifruit Leaf Disease Recognition

To identify the kiwifruit foliar disease diagnosis effectively and precisely,and to support methodologically the kiwifruit foliar disease diagnosis,the study on visual identification and kiwifruit foliar disease diagnosis with computer has significant effect and potential practical value on improving the reliability and efficiency of the kiwifruit foliar disease diagnosis,and on promoting the intelligence level of diagnosis of crop diseases.The research method of kiwifruit foliar disease automatic diagnosis,according to the mosaic,leaf blight,and ulcer as the research object,provides technical reference for the realization of automatic diagnosis of kiwifruit foliage disease.The work of this paper is as follows:(1)Under the condition of daily illumination,the images of kiwifruit leaf disease in kiwifruit garden have a complex background,which makes the common segmentation methods invalid.In order to solve this problem,this paper presents a scheme for complex background.This scheme through multiple morphological transformations to eliminate most of the background,the gray level of the background is 0,showing the original color image for target;and then use 2*R-G-B as the color factor Otsu threshold segmentation and L*a*b* color space of the K-means clustering algorithm,and proposes a segmentation algorithm for complex background image of Kiwi disease that can effectively segmentation of lesion,background and target segmentation is clear and pure.(2)Study on the extraction method of the characteristics of kiwifruit foliage disease lesion image.The extraction of kiwifruit leaf lesion characteristics of the image segmentation is an important step to realize the identification of kiwifruit foliage disease.Whether eye recognition or machine recognition,leaf lesion color and texture are an important basis for the classification of diseases.According to the characteristics of kiwifruit foliage disease lesion color,texture,structure and describes the proper parameters,extracted 18 characteristic value,and then use PCA(principal component analysis)on the characteristic parameters optimization,characteristic parameters optimized for 6.(3)Study on identification method of kiwifruit leaf disease.Using support vector machine and artificial neural network to classify the kiwifruit leaf disease with color features and texture features.The correct recognition rate of three diseases using artificial neural network are respectively 0.792,0.906,0.821;and the use of support vector machines to identify three kinds of diseases of the correct rate of 0.917,0.906,1.0.Compared with BP neural network,SVM performs better.Therefore,this paper uses support vector machine classification model to realize the design of kiwifruit disease recognition system.