2022 Fiscal Year Final Research Report
Machine learning-based noise filtering for rapid scan STEM image and its application to in-situ 3D dislocation observation
| Project/Area Number |
21K20491
|
|---|
| Research Category |
Grant-in-Aid for Research Activity Start-up
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
0401:Materials engineering, chemical engineering, and related fields
|
|---|
| Research Institution | Kyushu University |
|---|
Principal Investigator |
Ihara Shiro 九州大学, 先導物質化学研究所, 助教 (60909745)
|
|---|
| Project Period (FY) |
2021-08-30 – 2023-03-31
|
| Keywords | 走査透過電子顕微鏡法 / 深層学習 / その場観察 / 転位 |
|---|
| Outline of Final Research Achievements |
Fast imaging adaptable to in situ observation by using scanning transmission electron microscopy (STEM) yields STEM-specific noise and image distorion. This study developed the image distortion scheme and performed deep learning to restore the quality of STEM images acquired by fast scanning. The developed series of scheme was also utilized to in situ observation.
|
| Free Research Field |
金属材料
|
| Academic Significance and Societal Importance of the Research Achievements |
STEMはTEMで扱える範囲よりも厚い試料を観察可能である等，広く用いられているTEMよりも優れた点が多いものの，高速撮像には不向きであったことから，これまでその場観察に用いられることは少なかった．本研究で開発した手法によって，高速かつ高品質な像取得が可能となり，STEMを用いたその場観察をより実用化に近づけられた．今後，ナノスケールにおける動的な現象の解明に貢献し得ると考えられる．
|
