Computed tomography of sawlogs : knot detection and sawing optimization

Branches on trees introduce defects on sawn timber called knots. By scanning sawlogs using computed tomography, knots can be detected and accounted for so that the sawing process can be optimized with respect to outgoing product value. How the optimization should be done differs depending on available sawing equipment and the production strategy of the sawmill. It is important to investigate interesting production strategies with computer simulations to obtain an approximation of the profitability for a sawmill if investing in a computed tomography scanner. Another important step in the optimization process is to automatically segment knots so that they can be used by a computer when optimizing. This thesis presents an algorithm that automatically segments knots from computed tomography images of logs. The algorithm uses variable thresholds to segment knots on cylindrical shells of the computed tomography images. The knots are fitted to ellipses and matched between several cylindrical shells. The algorithm was tested on a variety of Scandinavian Scots pine (Pinus sylvestris L.) and Norway spruce (Picea abies (L.) Karst.) with a knot detection rate of 88-94 % and generating about 1 % falsely detected knots.Knots are defects with high impact on boards that are graded with respect to their bending strength. Some sawmills specialize in the production of such boards and this thesis includes a simulation study of sawing Norway spruce (Picea abies (L.) Karst.) logs to optimize the outgoing board value for such a sawmill. The production strategy investigated in this thesis was scanning of sawlogs with computed tomography and optimizing the rotational positioning of the logs in the sawing process. This study showed a possible mean value increase of the sawn timber by 11 %.There are additional degrees of freedom in log breakdown than rotational positioning, such as log spatial position, skew and which sawing pattern to use. If every possible combination of sawing parameters would be simulated, enormous computational resources would be required. A study made in this thesis investigates the feasibility to use only parts of the knot information when optimizing log rotational position. This is done by projecting all knots to a plane perpendicular to the log lengthwise direction and filter out the least significant knots. The study showed a great challenge in this approach and the presented algorithm was insufficient in its present form to compete with alternatives that use full information of the knots.Godkänd; 2013; 20130827 (erikjo); Tillkännagivande licentiatseminarium 2013-09-24 Nedanstående person kommer att hålla licentiatseminarium för avläggande av teknologie licentiatexamen. Namn: Erik Johansson Ämne: Träteknik/Wood Technology Uppsats: Computed Tomography of Sawlogs – Knot Detection and Sawing Optimization Examinator: Professor Anders Grönlund, Institutionen för teknikvetenskap och matematik, Luleå tekniska universitet Diskutant: Professor Johan Carlson, Institutionen för system- och rymdteknik, Luleå tekniska universitet Tid: Fredag den 18 oktober 2013 kl 10.00 Plats: Hörsal A, Luleå tekniska universitet, campus Skellefte