Wood Classification Based on Fiber Texture Using Backpropagation Method

Indonesia is one of the countries with abundant biological wealth, 125.9 million ha is the forest area in this country. Each type of wood has a different level of quality. Usually in one type of wood is divided into several classes to classify the quality. One type of grouping is through the texture of wood fiber. In the wood industry in Indonesia, the process of sorting wood classes is done by manual visual observation by an expert. This manual observation results in inconsistent product quality, long time, and more expensive costs. With these problems, a system is designed to classify wood classes based on the texture of the fiber. The type of wood that will be used is Cedar wood. Cedarwood is one type of wood that has very good quality. Its characteristic is to have a strong aroma produced from resin, not easily cracked when formed and has a high level of resistance in all conditions. Cedar wood will be grouped into 3 classes. This study will conduct a classification process of wood class based on the texture of the fiber using the Backpropagation and HOG methods. From the system that has been made produces an accuracy rate of 90%.