Penggabungan ciri morfologi, tekstur, dan bentuk untuk identifikasi daun menggunakan Probabilistic Neural Network

Abstract

This research proposes a new system for leaf identification using classifier combination and probabilistic neural network. The features that used for classification are morphology, texture, and shape of leaves. Classifiers combination is used to combine these features. The method to extract texture and shape are co-occurrence matrix and Fourier descriptors, respectively. After feature extraction, the feature is classified using probabilistic neural network. The experiment data consist of thirty species flora from Bogor Botanical Garden, Indonesia. The experiment results show that classification without classifier combination has the accuracy of 79.05%, and using classifier combination the accuracy increased to 83.33%. Hence, the propose system is promising for leaf identification and supporting plant biodiversity in Indonesia..