Determining a maximum value yield of a log using an optical log scanner

Two pattern recognition algorithms are proposed to detect two types of knots in lumber, lumped and surface knots, using an optical scanner. It is shown that lumped knots can be recognized by a log profile approach and surface knots can be detected by an image processing approach. A mathematical model for describing the geometry of the knots inside the log based on the surface parameters of the knots is presented. By a log breakdown strategy, the log can be cut into a maximum number of boards with prescribed dimensionalities. The mathematical knot model can be used to predict the locations, sizes, orientations, and shapes of the knots on these boards, whereby their lumber grades can be determined. Summing up the value yields of all the boards gives the value yield of the log. By computing all the possible value yields of the log from different log orientations, one can determine a maximum value yield of a log