2 Level Simplified Fuzzy ARTMAP for Grape Leaf Disease System Using Color Imagery and Gray Level Co-Occurrence Matrix

Abstract

This study presents a 2-Level Simplified Fuzzy ARTMAP (2L-SFAM) for classifying and recognizing images of grape leaf disease. The 2L-SFAM has been developed to extend the Simplified-Fuzzy ARTMAP (SFAM) to make it more suitable in particular applications. This proposed 2L-SFAM network uses multi - vigilance parameters which can be applied for classifying on data with 2 different patterns within the same category, for example grape diseases with multiple stages of each disease. In this work, color imagery and gray level co-occurrence matrix (GLCM) are used to classify attribute of data. Self-Organizing Feature Map (SOFM) is then used to extract disease region of grape leaf. The classification and recognition of grape leaf disease area are performed by 2L-SF AM. The main advantage of 2L-SF AM is the capability to learn totally new category of data without retraining the whole network and incrementally update the learned category data, which can reduce time for learning and classifying efficiently. The performance of proposed algorithm shows desirable accuracy in which the 2L-SF AM can efficiently classify and recognize diseases of grape leaf and stages of disease.