Plant Leaf Disease Diagnosis from Color Imagery Using Co-Occurrence Matrix and Artificial Intelligence System

Abstract

This paper presents plant leaf disease diagnosis system from color imagery using unsupervised neural network. Images are processed using both color and texture features. The system is mainly composed of two processes: disease feature extraction and disease classification. The process of disease feature extraction analyzes feature appearance using statistic-based gray-level co-occurrence matrix and texture feature equations. The disease classification process deploys the unsupervised simplified fuzzy ARTMAP neural network to categorize types of disease. Four types of grape leaf disease images are used to test the system's classification performance which are rust, scab, downy mildew and no disease. Desirable results have been achieved with more than 90% of accuracy. The proposed system can be applied to diagnosis other type of plant disease sufficiently.