Abstract
Atomic force microscopy (AFM) enables direct visualisation of surface topography at the nanoscale. However, post-processing is generally required to obtain accurate, precise, and reliable AFM images owing to the presence of image artefacts. In this study, we compared and analysed state-of-the-art deep learning models, namely MPRNet, HINet, Uformer, and Restormer, with respect to denoising AFM images containing four types of noise. Specifically, these algorithms’ denoising performance and inference time on AFM images were compared with those of conventional methods and previous studies. Through a comparative analysis, we found that the most efficient and the most effective models were Restormer and HINet, respectively. The code, models, and data used in this work are available at https://github.com/hoichanjung/AFM_Image_Denoising.
Original language | English |
---|---|
Article number | 103332 |
Journal | Micron |
Volume | 161 |
DOIs | |
Publication status | Published - 2022 Oct |
Bibliographical note
Funding Information:This research was supported by Brain Korea 21 FOUR . This research was also supported by Korea Institute for Advancement of Technology (KIAT) grant funded by the Korea Government (MOTIE) ( P0008691 , The Competency Development Program for Industry Specialist).
Publisher Copyright:
© 2022 Elsevier Ltd
Keywords
- Atomic force microscopy image
- Deep neural network
- Image denoising
- Image restoration
ASJC Scopus subject areas
- Structural Biology
- General Materials Science
- General Physics and Astronomy
- Cell Biology