Basil leaves disease classification and identification by incorporating survival of fittest approach

Abstract

Basil leaves are the major ingredients in traditional medicinal drugs with significant health benefiting phytonutrients. The quality of leaves describe the measure of excellence from deficits, substantial variations and defects. Also, the disease on leaves poses threats to the economic and production status in the agricultural industry worldwide. So, a new classification model using survival of fittest approach with successive generation of best results is proposed in this study. A three level hierarchical approach is developed to recognize basil leave diseases. Healthy and diseased leave samples with downy mildew and cercospora leave spot are taken as experimental objects. For enhancing contrast, a Contrast Limited Adaptive Histogram Equalization algorithm is performed by adjusting intensities of the image in order to highlights the target area for the segmentation of the disease from their background. Experiments are performed by utilising combination of texture and color features. Afterwards Random Forest, feature selection technique is used to attain the high informative features. Finally, a new approach of classification is employed to characterize the leave diseases. Our new classification model effectively detects, classified basil leaves diseases with 95.73% accuracy. Proposed model attain the maximum success rate than other existing methods.