Fruit Disease Classification based on Texture Features

Nowadays digitization and automation of machine in agriculture field plays prominent role. In this paper, we have proposed method to classify fruit as diseased and non-diseased. Firstly, we used K means clustering method for segmentation of diseased regions. Later, we used to extract texture features on segmented diseased regions. We have collected fruit diseased images from internet to create dataset and totally we have collect 2500 images from 10 fruit classes. We have conducted extensive experimentation using Probabilistic Neural Network and results shows that proposed method gives better performance compared to SVM and KNN.