Support Vector Machine Modeling For Identification Of Ferns Plant Based On Spores Image In Eigen Space

Abstract

Spores of is a breeding tool of ferns plant (Pteridophyta), which generally exist beneath the surface of each leaf. The research of Pteridophyta spores characteristic which commonly done is observing the size and shape to see what kind of spores. The method requires a lot of comprehension, experience, accuracy and time to achieve high accuracy in the determination of the type. Based on this reason, it is necessary to develop another technique by modeling to help identify the type Pteridophyta on the spore part. The purpose of this study is to make Support Vector Machine (SVM) modeling to identify the types of ferns by spore image in the eigen space. The data used was the image of the spores with four classes, each has 24 data per class. The method used in this study was the 2D-PCA for feature extraction and SVM for classification of data by using RBF and polynomial kernel functions. The result of this study is the accuracy of any extraction that has been attempted. The best result lies in the RBF kernel function with feature extraction of 70% and parameter 10 with an accuracy of 98.96%.