Classification of Tomato Leaf Based on Gabor Filter Extraction And Support Vector Machine Algorithm

Abstract

Tomato production in Indonesia is reduced because tomato leaves are stricken with disease. The main disease that often attacks tomato leaves is rotten leaves and bacterial patches or commonly called dry patches. Identification of tomato leaf disease is still done manually with human vision. The shortcomings of the method manually required a technology that is able to extract the texture of tomato leaf disease. One of them is by the process of extracting the texture of leaves with gabor filters, namely by using frequency and orientation parameters. Based on the results of the experiment obtained that the input parameter gabor filter with orientation of 90^o with a combination of frequency 4 produces a fairly clear contrast. The process of extracting the texture of the leaf aims to get the magnitude value of the tomato leaf that will be used as inputs for the classification process. The svm algorithm grouped data that had the same characteristics into one class. Training data used 42 images and test data used 30 images, with the success rate of 83.33%.