Statistical properties of lines distribution for tropical wood recognition system

Abstract

An automated tropical wood recognition system has been developed by Centre for Artificial Intelligence & Robotics (CAIRO), Universiti Teknologi Malaysia (UTM) based on machine vision to emulate the experts known as KenalKayu. The system used Statistical Properties of Pores Distribution (SPPD) feature extractor and K-Nearest Neighbor (KNN) classifier which have been proven to increase the system’s accuracy. Unfortunately, when more wood species were added to the system’s database, it reduces the accuracy of the system. Therefore, providing additional features that are representation of each species is one way to improve this issue. As the wood surface pattern is not only defined by pores but lines as well, this thesis presents additional new feature extraction method based on Statistical Properties of Line Distribution (SPLD) to capture the discriminant line features of each species and K-Nearest Neighbor (KNN) is used as classifier. The results of experiments showed that when used by itself as a feature extractor, the SPLD managed to achieve 88% accuracy, and the accuracy increased to 99.5% when combined with SPPD features and 100% accuracy was achieved when SPPD and Basic Grey Level Aura Matrix (BGLAM) features were used in combination. In conclusion, the SPLD method is an essential customized feature extractor and could be used as an alternative to adequately identify wood species. Hence, in the future, other discriminant features can also be added for wood identification purposes.