Image Pattem Classification for Plant Disease Identification using Local Tri-directional Features

Abstract

Plant diseases are main culprits of reduction in production in agronomy, which inflicts huge economic damages. Efficient detection and classification of plant diseases contributes significantly towards reducing these losses. This paper presents computer vision framework for plant disease identification and classification. The proposed system extracts Local Tri-directional Patterns (LTriDP) from plant leaf images of different classes. LTriDP features efficiently extracts discriminant information and represent each class with reduced dimensions. Classification is performed through multiclass support vector machines (SVM). Experiments are conducted on Tomato leaf dataset comprising five different classes. Experimental results show that the proposed framework out performs other methods based on commonly used feature descriptors and achieved 94% overall accuracy.