Identification Of Medicinal Plant Based On Fractal By Using Clustering Fuzzy C-Means

Abstract

Identification of medicinal plant species automatically is still a problem in recognizing various kind of medicinal plants in Indonesia. The research purpose is to develop medicinal plant identification system by using fractal and Fuzzy C-Means (FCM) clustering. Fractal method is used to extract image feature of the medicinal plant leaf However, FCM clustering is needed to classify image feature medicinal plant into some class or cluster. Two fractal approaches used in this research are fractal dimension and fractal code. Fractal dimension is based on similarity of the medicinal plant leaf image pattern and counted by box counting method, however fractal code is based on similiarity of the medicinal plant leaf image texture. This research used data from the collection in Biofarmaka !PB plantation and glasshouse of Ex-Situ conservation of Indonesian Tropical Forest. 600 data are used in this research consists of 20 variety of medicinal plants for each 30 samples. This experimental result shows that 85.04% of FCM clustering is based on fractal dimension and 79.94% FCM clustering is based on the fractal code.