Extraction and Exploratory Analysis of Texture Features on Images of Timber Defect

This paper discusses the development and analysis of texture features based on orientation independent Grey Level Dependence Matrix (GLDM) to characterise images of timber defects and clear wood. A series of processes including feature extraction, analysis of quantisation and displacement parameters, and feature analysis were implemented to facilitate data understanding in order to construct a good feature set that could significantly discriminate between defects and clear wood classes. To evaluate the discrimination capability of the features extracted, visual exploratory analysis was performed on defect and clear wood images of Meranti (Shorea spp.) timber species. Results from the analysis showed that there was substantial difference between defect classes and clear wood using the proposed multivariate texture features.