Hybrid approach for anthracnose detection using intensity and size features

Anthracnose is a group of fungal diseases that affect a variety of plants in warm, humid areas. Commonly infecting the developing shoots and leaves, anthracnose fungi (usually Colletotrichum or Gloeosporium) produce spores in tiny, sunken, saucer-shaped fruiting bodies known as acervuli. Symptoms include sunken spots or lesions (blight) of various colors in leaves, stems, fruits, or flowers. This reduces the agricultural production significantly. Hence there is a need to detect this disease at the earliest. This paper is based on the determination of lesion areas affected by anthracnose by segmenting them using thresholding segmentation technique. The lesion areas are identified by dark spots on the surface of the leaf or the fruit. The percentage of affected area is determined by taking the size and intensity of the spots into consideration. We have considered three types of fruit namely mango, apple and tomato for our work. The developed methodology consists of two phases. In the first phase, the segmentation technique namely region growing is employed for separating anthracnose affected lesion areas from normal area. In the second phase, these affected areas are graded by calculating the percentage of affected area The work finds application in developing a machine vision system in horticulture field.