Quantitative analysis of wood defect based on 3D scanning technique
Zhao Peng,Zhao Yun,Chen Guangsheng
DOI: https://doi.org/10.11975/j.issn.1002-6819.2017.07.022
2017-01-01
Abstract:Wood quality detection is a key issue in the wood manufacture factory or wood trade process. It consists of wood species recognition, wood physical parameter (such as density, hardness, water ratio, degree of surface roughness) prediction and wood defect detection, which are intimately connected with the efficient wood utilizations and wood prices. In the wood defect detection, the internal and external defects were inspected and processed with different schemes. It was an important way for effective wood grading and wood utilization to make the wood defect detection. In this paper, a detection and quantification scheme of wood defect was proposed based on three-dimensional (3D) laser scanning point cloud. This scheme could be used in the wood external defect detection such as cavity or tunnel. First, the Artec 3D Scanner was used to scan the wood surface to get the 3D point cloud. After preprocessing, theZ-axis coordinate value of current point was compared with the set threshold to judge whether it was a defect point. Second, a deep preferred search algorithm was used to classify the retained defect points marked with different colors. After this step, the segmented defects could be viewed with the Artec Cyclone software. Last, the integration algorithm was used to calculate the surface area and volume of every defect. In this step, every defect point was extended into a regular hexagon and a prism for the subsequent area and volume calculation by using the standard mathematical equations. The overall area or volume of every defect was computed by summarizing every defect point's area or volume. One detection system was realized with Visual C++ programming tool, the Artec 3D Scanner and a laptop. The simulation experimental results indicated that our scheme could accurately measure the surface areas and volumes of cavity or tunnel on wood surface with measurement error of 5%, if the defect's depth was less than 3 mm. This scheme could give the quantitative proofs for the subsequent wood grading and wood price. In fact, every 3D data point's format was (X,Y,Z,R,G,B,S), in which theR,G,B andS represented the red, green, blue and reflection information, respectively. Therefore, we could use theR,G andB information to perform the color classification for the wood surface by use of color moments or fuzzy classification algorithms. However, the wood defect points should be deleted in color classification in order to overcome the disturbance from wood surface's defect points. Fortunately, the deletion of defect points could be easily performed by use of our scheme, which was the advantage of our scheme compared to other wood parameter detection methods. Moreover, the used Artec Scanner was portable with small mass and volume (i.e. with a standard mass of 0.85 kg, a 3D scanning resolution of 0.5 mm, a size of 261 mm×158 mm×64 mm, and multiple data storage formats), so it could form a portable wood defect detection system with a laptop. In the future, with the development of 3D scanning instrumentation, the used 3D scanner can become more accurate with cheaper price, so our scheme may be conveniently used in wood manufacture factory or wood trade.