Light-weight single image super-resolution for embedded systems
We propose a learning-based single image super-resolution approach that is lightweight and suitable for embedded systems. In this project, we propose a texture extraction technique based on self-guided filter and residual interpolation. This texture extraction technique is used in the proposed super...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/74068 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | We propose a learning-based single image super-resolution approach that is lightweight and suitable for embedded systems. In this project, we propose a texture extraction technique based on self-guided filter and residual interpolation. This texture extraction technique is used in the proposed super-resolution method for better texture extraction for training low resolution to high resolution transformation. We further show how the proposed method is comparable if not better than current methods. Hardware acceleration and optimization are discussed to further improve performance and reduce computation complexity for implementation on embedded systems. |
|---|