Ekstraksi Daun Menggunakan Dimensi Fraktal untuk Identifikasi Tumbuhan Obat di Indonesia

Abstract

Identification medicinal plant species automatically still be a problem for recognizing various kind of medicinal plants in Indonesia. This research proposes a method for identifying a medicinal plant using Fractal Dimension and Probabilistic Neural Network (PNN). Fractal geometry offers an approach for the extraction of image which has the advantages in describing natural objects like leaves of medicinal plants. This is different from Euclidean geometry approaches that have been used to measure regular and symmetrical objects while the objects around us have an irregular shape. In this research, fractal dimension measured by Box Counting method. Fractal dimension has non-integer value that represent the self similarity of fractal. Identification results using Probabilistic Neural Network produces an accuracy of 67% and the results with showing the three classes that have the highest probability has the accuracy of 86.19%. Hence, the propose system is promising for medicinal plants identification in Indonesia.