Penggabungan Fitur Fuzzy Local Binary Pattern dan Fuzzy Color Histogram untuk Aplikasi Mobile Identifikasi Citra Tumbuhan Obat.

Abstract

This research develops a mobile application based on Android operating system for identifying medicinal plant images. The identification is conducted based on texture and color feature. This research uses medicinal plants leaf images. The total medicinal plants used in this research is 51 species taken from Biofarmaka IPB, Cikabayan Farm, Green house Center Ex-Situ Conservation of Medicinal Plant Indonesia Tropical Forest and Kebun Raya Bogor. Each species consist of 48 images, thus the total image used in this research is 2448. This research investigates effectiveness of the fusion between the Fuzzy Local Binary Pattern (FLBP) and the Fuzzy Color Histogram (FCH) in order to identify medicinal plants. The FLBP method is used for extracting medicinal plants texture. This method extends the Local Binary Pattern (LBP) approach by employing fuzzy logic for representing the local patterns of texture images. Fuzzification allows the distribution of the LBP values is used as a feature vector. Moreover, the FCH method is used for extracting medicinal plants color. This method considers the color similarity of each color pixel associated with all the histogram bins through fuzzy-set membership function. Next, Fuzzy C-Means (FCM) algorithm is used to compute the membership values. The fusion of FLBP and FCH is done by using Product Decision Rules (PDR) method. This research uses Probabilistic Neural Network (PNN) method to classify the FLBP and FCH feature vector. The experimental results show that the fusion between FLBP and FCH can improve the accuracy of medicinal plants identification (FLBP: 59.61%, FCH: 50.78%, fusion of FLBP and FCH: 74.51%).