Diagnosis of Powdery Mildew and Anthracnose cucumber Fungal Disease Using Image Processing and Artificial Neural Networks

Subject Areas :

Received: 2015-12-06 Accepted : 2016-01-23 Published : 2016-03-21

Keywords: Neural network, Segmentation, Anthracnose, Powdery Mildew, Feature extraction,

Abstract :

Given that most fungal diseases powdery mildew and anthracnose, cause to createdamage in cucumber greenhouses. In this study with a method of non-destructive based onimage processing and artificial neural network to distinguish the two types of fungal diseaseare discussed. Implementation Steps method proposed include three-part: segmentation,separate the damaged leaves and classification of the disease. After extracting the color andfeature from cucumber leaf. For separation different classes of image, samples will train bymulti-layer neural network by algorithm back-propagation error method. Input was Averagemain color components (R, G, B) images and 0 output as healthy leaf, 1 (powdery mildew)and two is Anthracnose. The structure of this network is 27-7-7-3. Use tansig transfer forhidden layer and output and between of educational functions, Levenberg-Marquardt’s wasthe most appropriate performance and accurate diagnosis can be 99/98 percent.