Identifikasi daun Shorea dengan Backpropagation Neural Network menggunakan ekstraksi fitur Discrete Wavelet Transform dan Ekstraksi Warna HSV

Abstract

Dipterocarpaceae is a major timber tree of tropical rain forest. Shorea is a genus of the Dipterocarpaceae family which consists of around 194 species. Shorea is difficult to be identified because it has a lot of diversity. It takes knowledge from an expert in the field of Shorea to be able to identify the types of Shorea. Errors in identifying the type of Shorea wood can lead to inappropriate selection for the final usability. In this research, we perform identification to 10 species of Shorea from Bogor Botanical Garden using Discrete Wavelet Transform and HSV color extraction as the feature extraction methods. Backpropagation Neural Network is used as the classification technique. The results of this research using both the DWT Haar family and HSV for feature extraction, and the DWT Daubechies 2 family and HSV produce 90% accuracy. The combination between DWT Haar family, DWT Daubechies 2 family and HSV color extraction produces 93.33% accuracy. The conclusion from the results of this research is the significant effect of HSV color extraction in increasing the accuracy for the identification of Shorea leaves.