Using geometric processing in the visualization of ring features in tomographic images of wood

We have developed a texture-based ring tracing algorithm to allow region-of-interest to be selected in terms of growth ring geometry during interactive visualization of 3D log data, as opposed to traditional way of selecting by voxel values alone. To pick out a single ring structure, we use a Fourier texture analyser to convert an image into a vector field so that a curve can be traced from any starting point easily. The extracted curve is thickened with the morphological distance transform. The thickened curve is used as a mask to filter the log data spatially and the result sent to a standard volume renderer for visualization. This allows a small region around a growth ring to be visualized, which is crucial to identifying certain defects in wood. The texture-based approach is more immune to noise, is better at discriminating non-ring features, and is more suitable for single curve tracing than traditional edge detection methods.