Automated cavity measurement using a region-growing algorithm

Abstract A digital image-processing technique based on a region-growing algorithm for cavity recognition was applied to micrographs of irradiation-induced voids and grain boundary cavities. As long as the resolution of the digitizing process was adequate, cavities were detected with at least an 80% efficiency for applications where illumination, low contrast, diffuse boundaries, and cavity internal structure made conventional analysis difficult. The accuracy of both manual and computer-based size measurements depended on how well boundaries could be defined. Although more work is needed to ensure reliable measurements for the range of conditions expected in materials characterization studies, this technique looks promising as the basis of an automated system for cavity analysis.