Defect Detection of Lumber Including Knots Using Bending Deflection Curve
Hiroaki NAGAI , Koji MURATA and Masashi NAKAMURA
Abstract:The commercial grading system for the strength of lumber is divided into visual grading and mechanical grading. Mechanical stress grading can estimate strength more precisely than visual grading, but the mechanical stress method cannot detect local defects such as knots that seriously affect mechanical strength. This study investigated methods of automatically detecting these knots at the time of measuring Youngfs modulus (E) with a grading machine. The results were as follows. (1) Edgeline of the lumber is extracted from a digital image. Bending deflection curve calculated from the displacement of the edgeline is measured using the digital image correlation technique. (2) Differential profile of the bending deflection curve around knots showed a characteristic peak and the profile pattern depended on the structure of knots, I.e. intergrown knot, encased knot and direction of spike knot. (3) The specimens, with a differential profile that showed a clear peak around knots, also ruptured at the knots. However, other specimens, with a differential profile that did not show a clear peak around knots, did not always rupture at the knot, although many of these specimens ruptured at the area without knots. Therefore, the method using a differential profile of bending deflection curve can distinguish two types of knots that would be evaluated as the same type by visual grading. Key Words:Bending deflection, Defect detection, Grading, Knot, Image Analysis