Cotton Leaf Disease Detection Using Texture and Gradient Features

Abstract

The detection of cotton leaf disease is a very important factor to prevent serious outbreak. Most cotton diseases are caused by fungi, bacteria, and insects. A new method is proposed for careful detection of diseases and timely handling to prevent the crops from heavy losses. A disease due to bacteria, insects and fungus occurs in the cotton leaves in the range of about 80-95%. In the proposed work, first the group of infected leaves and normal leaves are collected and the image preprocessing is done using Adaptive histogram equalization for enhancing the contrast. In feature extraction phase, texture and gradient feature are extracted using Local Binary Pattern (LBP), Histogram of Oriented Gradient (HOG) and Differential of Gaussian (DOG). K- Nearest neighbor classifier is applied to classify the leaf image as a unaffected or an affected leaf. A cotton leaf database is internally created to evaluate the efficacy of our algorithm. The validate results show that the proposed method achieved higher classification accuracy in lower computational time.