Geometrical analysis and pattern recognition using mapping technologies of three-dimensional fracture surfaces in materials

Geometrical analysis of fracture surfaces in materials was made using newly developed computer programs on the three-dimensional images reconstructed by the stereo matching method. The global value of the fractal dimension of the fracture surface was estimated by the box-counting method on a fatigue fracture surface of a Cu-Be alloy and impact fracture surfaces of a SiC and an alumina. The results of the present analysis were well correlated with those of the two-dimensional fractal analysis. The fractal dimension map (FDM) by the box-counting method and the surface roughness map (SRM) proposed in this study can give important information about the local fracture mechanisms, the crack growth direction or the fracture origin in materials. FDM and SRM have interesting characteristics by which one can discriminate the flat regions, the regions of complex geometry or the steeply inclined areas on a given fracture surface. Pattern recognition using mapping technologies of FDM and SRM is also applicable to the extraction of “hidden patterns” on fracture surfaces, which cannot be observed only by microscopes. C 2005 Springer Science + Business Media, Inc