Denoising of series electron holograms using tensor decomposition.

Area of Science:

• Materials Science
• Physics
• Data Science

Background:

• Low-dose electron holography is crucial for imaging beam-sensitive materials.
• Poisson noise in electron holograms degrades image quality and limits analysis.
• Existing noise reduction methods may not be sufficient for complex datasets.

Purpose of the Study:

• To develop and validate a novel noise-reduction technique for series low-dose electron holograms.
• To enhance the signal-to-noise ratio (SNR) and improve the accuracy of reconstructed phase maps.
• To demonstrate the applicability of the method to time-resolved in situ electron holography.

Main Methods:

• A third-order tensor representation of stacked 2D holograms with Poisson noise.
• Tensor decomposition into a core tensor and three factor matrices.
• Approximation to a lower-rank tensor using noise-free principal components.

Main Results:

• Significant improvement in peak signal-to-noise ratios for holograms and reconstructed phase maps.
• Successful application to simulated p-n junction in a semiconductor sample.
• Demonstrated effectiveness for time-resolved in situ electron holography with non-uniform fringe contrast and drift.

Conclusions:

• Tensor decomposition is an effective noise-reduction technique for low-dose electron holography.
• The method enhances quantitative phase reconstruction accuracy and precision.
• The approach is suitable for in situ experiments and beam-sensitive materials analysis.