CASR: A Robust Cyclic Framework for Arbitrary Large-Scale Super-Resolution with Distribution Alignment and Self-Similarity Awareness
Wenhao Guo, Zhaoran Zhao, Peng Lu, Sheng Li, Qian Qiao, RuiDe Li, [6pt] School of Artificial Intelligence, Beijing University of Posts, Telecommunications, Beijing, China · 2026
Read CASR: A Robust Cyclic Framework for Arbitrary Large-Scale Super-Resolution with Distribution Alignment and Self-Similarity Awareness by Wenhao Guo, Zhaoran Zhao, Peng Lu, Sheng Li, Qian Qiao, RuiDe Li, [6pt] School of Artificial Intelligence, Beijing University of Posts, Telecommunications, Beijing, China (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-22159.txt.
Description
Upscaling an image far beyond its training range usually falls apart: noise, blur, and artifacts pile up as the scale grows. Wenhao Guo and five co-authors reframe extreme magnification as a chain of small, in-distribution scale steps, in a framework called CASR. One single model walks the image up through those steps, keeping inference stable at arbitrary scales.
Two modules handle the hard parts: one aligns structural distributions across steps to stop error buildup, the other restores fine textures while preserving self-similarity. The design tackles distribution drift and patch inconsistencies together. Despite using only one model, CASR cuts distribution shift sharply, giving super-resolution a steadier ladder to climb.
Catalog record
| CREATOR | Wenhao Guo, Zhaoran Zhao, Peng Lu, Sheng Li, Qian Qiao, RuiDe Li, [6pt] School of Artificial Intelligence, Beijing University of Posts, Telecommunications, Beijing, China |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-22159