Cross-Correlation for Automated Stitching of Two-Dimensional Multi-Tile Electron Backscatter Diffraction Data (Preprint)

Abstract : A method for automatically aligning consecutive datasets of large, two-dimensional multi-tile electron backscatter diffraction (EBSD) scans with high accuracy was developed. The method involves first locating grain and phase boundaries within search regions containing overlapping data in adjacent scan tiles, and subsequently using cross-correlation algorithms to determine the relative position of the individual scan tiles which maximizes the fraction of overlapping boundaries. Savitzky-Golay filtering in two dimensions was used to estimate the background, which was then subtracted from the cross-correlation to enhance the peak signal in samples with a high density of interfaces. The technique was demonstrated on datasets with a range of grain boundary densities. The equations were implemented as enhancements to a recently published open source code for automated stitching of multi-tile datasets.