Classification of Japanese Fagaceae Wood Based on Microscopic Image Analysis

This paper presents wood identification method on microscopic image of Japanese Fagaceae wood by applying Daubechies Wavelet (DW) and Local Binary Pattern (LBP) algorithms. The main idea for this identification is to extract as much as possible the characteristics of wood to improve accuracy. The color intensity of microscopic wood image can be one of the characteristics of wood. Therefore, before extracting features with DW and LBP, we add color extraction to separate the color intensity in the Red, Green, and Blue (RGB) channels from the image. In addition, we compared several values of the LBP parameter for kernel type original, var, uniform, and ror to find the optimal value. The results of this work performed the accuracy of wood identification on microscopic image of Japanese Fagaceae wood with the Support Vector Machine (SVM) classifier is obtained 95.2% by setting P and R parameters at 8 and 2. Moreover in this paper, the most optimal LBP parameter value occurs when P = 16 and R = 1 for the original kernel type, which reaches 100% accuracy.