Three-dimensional stacked filter: A non-linear filter for series images obtained using a transmission electron...

Area of Science:

• Materials Science
• Microscopy Techniques
• Image Processing

Background:

• Quantitative high-resolution transmission electron microscopy (HRTEM) faces challenges with low signal-to-noise ratio (SNR), particularly for beam-sensitive materials and dynamic studies.
• Existing de-noising methods often struggle to effectively enhance image quality without introducing artifacts.

Purpose of the Study:

• To introduce a novel three-dimensional stacked filter (3DSF) for de-noising HRTEM images.
• To improve SNR, reduce artifacts, and enhance computational efficiency in HRTEM image analysis.
• To demonstrate the filter's effectiveness on periodic structures and dynamic processes.

Main Methods:

• A novel non-linear filter, the three-dimensional stacked filter (3DSF), was developed.
• HRTEM images are stacked into a 3D data cube and processed using a 3D domain Wiener filter.
• The method was validated using simulated and experimental images of graphene and metal-organic frameworks (MOFs).

Main Results:

• The 3DSF accurately estimates noise power spectral density, outperforming traditional 2D Wiener filters.
• Significantly improved SNR, reduced artifacts, and enhanced computational efficiency were achieved.
• The filter successfully analyzed an ultra-low dose HRTEM image stack of MOF MIL-101, revealing subtle lattice shrinkage over 40 frames.

Conclusions:

• The 3DSF is a powerful de-noising tool for HRTEM, especially for low-dose and dynamic imaging.
• This technique enables more detailed characterization of beam-sensitive materials and dynamic phenomena.
• The 3DSF offers a significant advancement in quantitative HRTEM analysis.