The Application of RBF Neural Network in the Wood Defect Detection

Abstract

Wood defect is due to the physiological process, genetic factor or affected by the external environment in the growth period. These defects will reduce the utilization value of wood. However, it is very difficult to determine whether there are defects exist, and the degree of defects. Therefore, the effective detection of wood defect information is particularly important. A new wood defect detection method by using RBF neural network was proposed in this paper. The new RBF defect detection method can be divided into the following main steps: (1) Detect wood defects by using X-ray nondestructive testing technology. (2) Deal with defect images by using digital image processing technology. (3) Analyze the information of different defects, and extract the characteristic value of wood defects. (4) Then, the RBF neural network model was constructed. (5) Finally, the RBF neural network is trained with the known samples and simulated with the unknown samples. The experimental results shown that the RBF neural network method was effectively detect the two typical wood defects. This method provides an important theoretical basis to realize the wood defect automatic detection.