Guava Leaf Disease Detection Using Colour Region Segmentation and Circularity Value Techniques

Abstract

Workers in the agricultural sector face various dangers that affect human society's food security, such as climate change, animal grazing, plant diseases, and other hazards that are well known. Plant disease is one of the most serious problems since it not only causes massive waste of plants for human consumption, but it also has a significant impact on human society's health and the lives of farmers whose primary source of income is the production of healthy crops. If infections are not diagnosed early, they can spread diseases, like Alga Leaf Spot, Rust, and Whitefly, which can infect the entire farm. This project aims to implement the Image Processing Technique, for plant disease detection on guava leaves in the early stage. First, Histogram Equalization is used to improve the image contrast and quality. Then the Colour Segmentation Technique was applied to detect the disease area on the leaf based on its disease-type colour. Next, the Circularity Value Technique was applied to detect the shape of the disease on the leaf.  Finally, the disease will be detected and classified based on its colour and shape. From the analysis, it is found that the disease colour is 100% correctly detected, 53.33% detected its correct shape, and 53.33% detected the correct disease.