Improvement Coconut Wood Classification Based On Gray-Level Co-Occurrence Matrices And Minimum Redundancy Maximum Relevance

Coconut wood(Cocos nucifera) is used as alternative and complementary raw material for making house and furniture. Decorative style and material strength of coconut wood started gaining attention as an alternative material in the furniture industry. The needs of furniture industries to determine the quality of a good coconut wood to make good product quality, requiring control during the selection process until the end of the process material to be a ready-made product. Determining the level of quality(grading) visually for coconut wood needs to be created automatically, so that it can be used for the determination of material suitable for use as furniture or construction materials for buildings and reducing dependence on human grader. This research presents our experimental work on coconut wood quality classification using k-Nearest Neighbors(k-NN), Discriminant Analysis(DA), Na?ve Bayes(NB) and Decision Tree(DT). The Gray-Level Co-occurrence Matrix(GLCM) is used to extract the texture features of coconut wood images. Minimum Redundancy Maximum Relevance(m RMR) feature selection is used to improve accuracy of classifier. Experiment result shows that Decision Tree(DT) based on GLCM + m RMR gives the best accuracy rate at 84.12%, which is slightly better than 81.80% of KNN2 based on GLCM.