Wood Identification Type by Using Support Vector Machine – Based Image

Abstract

Wood identification type in Indonesia usually has been done manually. The identification is done by monitoring wood pore on wood longitudinal section using loupe or microscope with 10 times zooming. The minimum computerized methods has done, due to the lack of the research in this field and the difficulty to find wood databases. This research was aimed to classify four types of wood trade in Indonesia, by using support vector machine based on image data. We used two-dimensional principal component analysis (2DPCA) methods for extracted image data. This methods can identify the type of wood quite fast, so it is speed up the identification of wood type. The best result of this researched has accuracy 95.83%. This proves that the methods used are suitable to be applied in identification of wood type.