Timber Identification Based on the Grain Structure Using Learning Algorithms

Lumber is used for many purposes all over the world. Grain structure is a characteristic of wood that can be used to identify different species. It refers to the pattern formed by the wood fibers as they grow. The grain can vary widely depending on the tree species, its growth conditions, and the part of the tree from which the wood is taken. Identifying the type of wood is really a difficult task for a normal person. A subject-matter expert is always expected for this purpose. Examine the grain pattern closely to compare it with known patterns of different species. While grain structure is a valuable tool for identification, it's important to consider other factors like color, texture, and density for a more accurate determination. The integration of these characteristics can enhance the precision of timber identification. This study aims to determine the type of wood through grain structure analysis. Deep learning algorithms can facilitate this achievement. A comparison of 3 different types of CNN model i.e., OwnNet, AlexNet and LeNet is done. LeNet model seems to produce the highest accuracy, So LeNet model is being used for processing. The cross-section of the wood is captured and uploaded in the web application. The data is processed using the classification-based algorithm (CNN - LeNet) on Neural Networks for analyzing and identifying the type of wood. Then the common name and scientific name are displayed immediately on the web application.