Developing Wood Identification System by Local Binary Pattern and Hough Transform Method

Abstract

The aim of this paper is to develop the wood identification system using two methods, called, Local Binary Pattern (LBP) and Hough Transform (HT). Here, 12 (twelve) varies of wood species form Indonesia will be used as the data sets. The wood species are taken from wood anatomy laboratory, Puslitbang Hasil Hutan (P3HH). Here, the classification method of this research uses Euclidean Distance (ED) to determine the distance of two images of wood. From the classification results, using LBP method is shown better than HT. The weakness of HT method in this paper is HT method can only detecting the circle shape rather than arbitrary shape. By using LBP method, 6 of 12 species are 100% accurate detected. Beside, using HT method, only one species (Cratoxylon formosum) has accuracy 90%.