A04 – Crystal Analysis of High Resolution Image Data

PI: Benjamin Berkels (AICES)

SFB researcher: Amel S. A. Alhassan (AICES)

The goal of this project is to create an unsupervised and robust way for quantitative defect characterization from high-resolution data at atomic scale originating either from experiments or simulations. The individual objectives are as follows.

• Robust characterization of the crystal unit cells as part of the segmentation,
• Quantification of the deviation of an individual crystal from the perfect lattice,
• Constructing a method for defect characterization of atomic resolution image data that simultaneously segments image data into crystals and quantifies deviations,
• Approaches for the comparison of defects in experimental and simulation results, and
• Theoretical foundation for the variational approaches, e.g., existence of minimizers.

To achieve the goal of characterizing defects in atomic resolution images of crystals, the task is modeled as two intertwined sub-problems: The segmentation of an image into individual grains or crystal types, and the quantification of the deviation of the single crystals from the perfect lattice. The final goal that is to develop a method able to solve the two sub-problems jointly since knowledge of the solution of one of the problems is beneficial for the solution of the other one.

At the end, the methods developed within project A04 will be used to create tools for the quantitative comparison of defects in experimental and simulation results.