Skip to main content
U.S. flag

An official website of the United States government

Progress in analysis of computed tomography (CT) images of hardwood logs for defect detection

Informally Refereed

Abstract

This paper addresses the problem of automatically detecting internal defects in logs using computed tomography (CT) images. The overall purpose is to assist in breakdown optimization. Several studies have shown that the commercial value of resulting boards can be increased substantially if defect locations are known in advance, and if this information is used to make sawing decisions. The problem is difficult, particularly for hardwood species, because of the natural variations of wood density and because of the irregular placement of defects such as knots. In our previous work, we developed a processing approach that utilizes artificial neural networks (ANN) to classify CT log images on a pixel-by-pixel basis. The system uses small (e.g., 5-by-5) neighborhoods in an image make a preliminary classification decision, using labels such as “knot,” “split,” and “bark.” This approach has yielded high accuracy statistically, with classification rates often exceeding 95%. Subjectively, however, the results can often be improved somewhat through further processing steps. For example, relatively simple operations can be employed to remove small. spurious regions that are not statistically significant, although their removal significantly improves the appearance of the results. This paper presents recent results for two hardwood species (red oak and sugar maple), as well as preliminary results for a softwood species (black spruce).

Citation

Sarigul, Erol; Abbott, A. Lynn; Schmoldt, Daniel L. 2003. Progress in analysis of computed tomography (CT) images of hardwood logs for defect detection. Proceedings, ScanTech 2003, The Tenth International Conference on Scanning Technology and Process Optimization in the Wood Industry. 19-30.
https://www.fs.usda.gov/research/treesearch/6140