A Layout for Citrus Plant Disease Detection Using Automation Tools

Abstract

The Computer Aided Diagnostic (CAD) tools have gained a significant position in the tools
required for improvement and expansion of quality in production yield by reducing manual error in
detection and diagnosis of disease. The CAD tools have properties of expediting the decision making
capability of the system on the complex disease data automatically. Since these plant diseases can be
fatal in terms of economic loss and nutrition loss, there is an increase in need of identification of
disease at early stage so as to meet nutritional needs of human beings. The use of these computer
aided methods has made it possible to reduce the production loss and early detection of diseases. This
article proposes a complete layout for detection of Citrus disease using image processing
methodology. The images acquired from the plant fields are processed in terms of noise removal and
contrast improvement in order to pass through subsequent stages accurately. The automation of
disease detection comprise of various stages namely-image acquisition, preprocessing, image
segmentation, feature extraction, feature selection and classification. For carrying out detection the
acquired image is generally taken as leaf or fruit as the major symptoms of disease appears on the
leaf and fruit portion. After image acquisition the image is pre-processed in terms of contrast
enhancement and color space conversion. The pre-processed image is passed through segmentation
stage for selection of region of interest and the features are extracted to form Image feature vector.
These feature vector act as training data for classifier. The results of the classifier are quantified in
terms of efficiency calculated through the values of true and false cases. This include True positive,
False positive, True negative and False negative