Automatic knot detection and measurements from X-ray CT images of wood: A review and validation of an improved algorithm on softwood samplesReport as inadecuate




Automatic knot detection and measurements from X-ray CT images of wood: A review and validation of an improved algorithm on softwood samples - Download this document for free, or read online. Document in PDF available to download.

* Corresponding author 1 LERFoB - Laboratoire d-Etudes des Ressources Forêt-Bois 2 ADAGIO - Applying Discrete Algorithms to Genomics and Imagery LORIA - ALGO - Department of Algorithms, Computation, Image and Geometry 3 LERMAB - Laboratoire d-Etude et de Recherche sur le Matériau Bois

Abstract : An algorithm to automatically detect and measure knots in CT images of softwood beams was developed. The algorithm is based on the use of 3D con- nex components and a 3D distance transform constituting a new approach for knot diameter measurements. The present work was undertaken with the objective to automatically and non-destructively extract the distributions of knot characteristics within trees. These data are valuable for further studies related to tree development and tree architecture, and could even contribute to satisfying the current demand for automatic species identification on the basis of CT images. A review of the literature about automatic knot detection in X-ray CT images is provided. Relatively few references give quantitatively accurate results of knot measurements i.e., not only knot localisation but knot size and incli- nation as well. The method was tested on a set of seven beams of Norway spruce and silver fir. The outputs were compared with manual measurements of knots performed on the same images. The results obtained are promising, with detection rates varying from 71 to 100%, depending on the beams, and no false alarms were reported. Particular attention was paid to the accuracy obtained for automatic measurements of knot size and inclination. Comparison with manual measurements led to a mean R2 of 0.86, 0.87, 0.59 and 0.86 for inclination, maximum diameter, length and volume, respectively.

Keywords : Branchiness 3D distance transform Computer tomography Picea abies Abies alba





Author: Fleur Longuetaud - Frédéric Mothe - Bertrand Kerautret - Adrien Krähenbühl - Laurent Hory - Jean Michel Leban - Isabelle Debl

Source: https://hal.archives-ouvertes.fr/



DOWNLOAD PDF




Related documents