Applicability of Motion Estimation Algorithms for an Automatic Detection of Spiral Grain in CT Cross-Section Images of Logs

摘要

Techniques for an automatic detection of spiral grain in cross-section CT images of logs are proposed and evaluated on sets of natural and artificial cross-section images. Explicit analysis of global rotation, block matching, and optical flow techniques are compared. Experimental results seem to indicate that spiral grain in fact cannot be modeled by a circular motion of luminance values in gray scale images.