Joint non-rigid image registration and reconstruction for quantitative atomic resolution scanning transmission...

Area of Science:

• Materials Science
• Microscopy
• Data Analysis

Background:

• Aberration-corrected scanning transmission electron microscopy (STEM) offers atomic resolution for analyzing materials.
• Scan artifacts, including noise and distortions, hinder precise quantitative analysis of STEM datasets.
• Accurate determination of local strain fields, composition, and bonding states is crucial for materials characterization.

Purpose of the Study:

• To develop a novel bias-corrected non-rigid registration approach for STEM image series.
• To compensate for both fast and slow scan artifacts that limit quantitative interpretation.
• To improve the precision of atomic column localization and subsequent data analysis.

Main Methods:

• A bias-corrected non-rigid registration algorithm was developed.
• The method explicitly couples deformations across individual images in a series.
• Minimization of average deformation was used to correct slow scan artifacts and reduce noise simultaneously.

Main Results:

• The novel approach effectively compensates for fast and slow scan artifacts in STEM image series.
• Simultaneous reduction of image noise and scan distortions was achieved.
• The method was validated using both synthetic and experimental STEM datasets.

Conclusions:

• The presented registration technique significantly enhances the quantitative interpretation of STEM datasets.
• It enables more accurate atomic resolution strain and elemental mapping.
• This advancement is vital for precise materials characterization at the nanoscale.