Tropical hardwood species identification based on first order statistical moment of cross section images color and texture extraction using gray level co-occurrence matrix

Hardwood is a raw material that is easily processed to be used as a non-oil valuable commodity, such as used in the furniture industry and building a residential house. Some species of hardwood used as raw material for the production of furniture are Teak (Tectona grandis), Mahogany (Swietenia mahagoni), Shorea (Shorea Sp.) and Philipines Mahogany (Pterocarpus indicus). These species of woods grow in most tropical forests in South and Southeast Asia countries such as India, Indonesia, Malaysia, and Philipines. The recognition process of these four types of wood is generally dependent on an individual's expertise in identifying the species of wood based on general characteristics (without a microscope) or anatomical (under a microscope). The recognition process is vulnerable since it relies on the subjectivity of experts and their inexperience in this field of work. In this study, images are captured and used to identify the type of hardwood. The identification is based on the general characteristics of hardwood in the texture and color using statistical methods for the extraction of color and statistical methods of second order Gray Level Co-Occurrence Matrix (GLCM) for feature extraction texture. Classification is done using Multilayer Backpropagation Neural Network. The results of experiments achieve 90% accuracy. The results obtained showing a high rate of accuracy proofs that the techniques used are suitable to be implemented for hardwood recognition system.