Matching Component Analysis for Microtexture Region Segmentation

Microtexture regions (MTR) are collections of grains with similar crystallographic orientation; their presence in aerospace components can significantly impact compoent life. Thus, a method to detect and characterize MTR is needed. A potential solution is to fuse information from multiple nondestructive evaluation techniques that are sensitive to local crystallographic changes in titianum. In this work, we introduce a technique that combines a variant of the matching component analysis algorithm with level-set inversion in order to determine MTR boundaries using multiple sets of nondestructive evaluation data. Although this method is agnostic to the type of nondestructive evaluation data, here it is used to combine information from multiple rotations of an asymmetric eddy current coil. The method is successfully demonstrated on simulated eddy current data of a real titainum specimen.