Recognition and classification of Produce affected by identically looking Powdery Mildew disease

Abstract

Abstract Plant diseases are visually observable patterns of a particular plant. Varieties of plant diseases, which are recognized by human beings, are identical or look similar in appearance. In this paper, we have considered recognition of fungal disease symptom like powdery mildew, looking similar in appearance affected on different produce. The powdery mildew symptom affected on produce like grape, mango, chili, wheat, beans and sunflower are considered for classification. Colour and texture features are extracted from image samples of produce affected by powdery mildew symptom. The extracted features are then used as inputs to knowledge-based and artificial neural network (ANN) classifiers and tests are performed to classify image samples. The colour analysis is done using Red, Green, Blue (RGB) and Hue, Saturation, Intensity (HSI) models. Texture analysis is done using gray level co-occurrence matrix (GLCM). The overall average classification accuracy with colour, texture and combined features are 70.48 %, 70.07 % and 76.61 % respectively using the ANN classifier. The overall average classification accuracy has increased to 71.92 %, 80.60 % and 87.80 % with colour, texture and combined features respectively using the knowledge-based classifier.