Tea Plant Leaf Disease Identification Using Hybrid Filter and Support Vector Machine Classifier Technique

Abstract

In our country, Agriculture is the major occupation of the people. If a plant affected by any disease in long time, then there is a shortage in productivity in agriculture. Therefore, it is essential to diagnose and analyze the infection. Tea leaf cultivation is highly labor intensive and provides employment to about 2.0 million families engaged in tea cultivation, trade and trade across India. During cultivation, tea is most affected by the disease. In this research, various diseases in the tea plants are studied and also with help of image processing techniques and pattern recognition techniques, diseases are recognized at early infected stage. The method presented here is to arrange the leaf spot, rhizome rot, powdery mildew diseases and leaf blotch diseases which are infected in the tea leaf plantation. The color transformed images are sharply segmented using watershed transformation algorithm. Multiclass SVM classifier classifies the tea leaf diseases using gradient feature values of the tea leaf images. In this paper, we used hybrid filter which comprise of median filter and Gaussian filter for the purpose of edge detection and noise reduction. Finally, the performance evaluated in terms of accuracy, and it is found that the presented system is realizable and provides better classification than earlier techniques.