Deep learning for magnetic domain image denoising and super-resolution
Project Description

Images obtained from microscopes may not have desired resolution due to the existence of various noise channels and imperfect microscope setup. We use deep learning and physics-based techniques to enhance the resolution of noisy magneto-optical Kerr images.

Supervisor
SHAO Qiming
Co-Supervisor
SHAO Qiming
LI, Xiaomeng
Quota
3
Course type
UROP1000
UROP1100
UROP2100
UROP3100
UROP4100
Applicant's Roles

The student will be working on the following tasks:
1. learn software PyTorch (deep learning tool), OpenCV (image processing tool), and mumax3 (micromagnetic simulation tool) through detailed tutorials
2. get familiar with MOKE imaging and generate ideal magnetic domain images
3. use deep learning to classify ideal images
4. understand noises in magneto-optical Kerr images
5. use deep learning to achieve super-resolution

Applicant's Learning Objectives

The student will learn image processing and deep learning techniques. Besides, the student will learn the physics behind microscope imaging and magnetic domain imaging.

Complexity of the project
Moderate