Paddy Crop Disease Detection using GLCM Feature Extraction and SVM Technique

Abstract

A plant disease detection method using images of infected leaves were taken to create a disease leaf database dataset and another healthy leaf dataset. Different image processing steps have been used, such as preprocessing, gray conversion, segmentation. The leaf detection testing scheme comprises of the same steps and the metrics acquired are compared to the current healthy and diseased leaves trained database. GLCM or gray Level Co-occurrence Matrix characteristics were assessed for a collection of disease information from the three common rice plant disease i.e. brown spot, tight brown spot and paddy blast disease. For each disease, the training function dataset was created using GLCM characteristics. Processing steps are introduced to a test picture at testing stage and the GLCM characteristics for this present test picture are assessed. Finally, leaf color texture classification is performed using multi- SVM classification to achieve feature extraction metrics for both healthy disease and diseased leaves.