Ancient characters
Character image understanding
Exploring recognition, enhancement, restoration, and visual analysis workflows for historical scripts and character images.
About
Hi, I am Wentao Ruan, also known as INWLY.
I am a member of the IIP-HCI Lab at Anhui University. My work sits between visual computing and practical systems: ancient character image processing, image restoration, generative models, and tools that make research easier to use.
Current directions
Research interests and experiments
Restoration
Degraded image repair
Building cleaner visual representations from degraded, noisy, incomplete, or low-quality image data.
Generation
Character generation systems
Studying how generation methods can support character synthesis, repair, and visual understanding.
Publications
CVPR Findings 2026
LWTformer: A Detail-Aware, Learnable Wavelet-Transformer for Ancient Chinese Character Image Restoration
LWTformer restores degraded ancient Chinese character images with learnable wavelet downsampling, spatial-enhanced attention, and wavelet-aware refinement for clearer stroke details.
Build log
Research notes, code notes, and small experiments.
This site is becoming a home for things I learn while moving between papers, datasets, code, and product-like prototypes.
Recent writing
Latest notes
Contact
