Image Analysis for Particle Tracking in Wood Cutting Processes
Austria is a key player in the forestry and wood industry. Separating or cutting of wood is one of the most important process steps resulting in huge amounts of particles of different sizes (chips, dust). The rapid removal of particles from the process zone and from the processing space is thus of great importance (surface quality, dust, health issues).
The problem of chip removal is already known, but up to now is mainly limited to trial and error. The high cutting speeds (up to 100 m / s) lead to extremely high acceleration of the separated particles. A purely visual observation of the particles, the analysis of the interaction of the particles with each other and the determination of the influence of the air flow, due to the extraction and the rapidly rotating tool, therefore is extremely difficult. In addition to these problems, the electrostatic charge of the particles makes it difficult to predict the behavior of the particle.
For this purpose, a high-speed camera with a correspondingly high frame rate and short exposure time is used. The frames of the recordings obtained are then analyzed by means of the so-called optical flow (Horn and Schunck 1981). Then, the trajectory of the individual particles can be traced in detail, and using the speed, particle size and mass, the kinetic energy can be calculated. Furthermore, the reflective properties of the particles and the loss of speed after the collision are analyzed using this method. For this purpose, it is necessary to take images from two different directions of view and to correlate the data of these two images using mathematical methods. This analysis provides statistical probabilities for the particle behavior depending on the particle’s size.
Mathematically, using image analysis, a method characterizing the movement and the amount of particles under simplified laboratory conditions has to be implemented. The transport equation is used to model the movement of the particles. To find the displacement vector (parameters) is the so-called inverse problem. The solution to this problem is not unique, in general. Therefore, one has to choose one (called regularized) solution that fits best to this problem. On the implementation level, one has to determine how the choice of regularization effects the result, and whether the use of alternative methods could bring an improvement.
To study and solve this problem, the Computational Science Center (CSC) , University of Vienna, (Prof. Otmar Scherzer and PhD st. Thomas Glatz) takes part in a project consisting of the Kompetenzzentrum Holz GmbH and international industry partners.
You must be logged in to post a comment.