Abstract

The recognition of defects in the wood manufacture process is ever more necessary and fundamental. However, the utilization of robust techniques that will allow these machines to provide a high degree of reliability is still a matter for study. Images obtained with X-rays and other techniques are already being used in the wood industry. However, the challenge remains to develop robust detection methods for different species and considering their distinct conditions. In this article, we present a defect segmentation method, initially developed to detect wood knots in Pinus radiata, based on images taken with X-ray computerized tomographies. This process uses deformable contour techniques (snakes), basing their evolution on a simulated annealing search method. The results obtained are highly satisfactory with respect to other approaches and the methodology appears to be a robust method of defect detection in round lumber under different conditions.