DMSH-Net: Depth-aware multi-scale hybrid vision network for image dehazing
Chenping Zhao, Jun Li, Yingjun Wang, Zuhua Guo, Xiaoyue Li · 2026
Read DMSH-Net: Depth-aware multi-scale hybrid vision network for image dehazing by Chenping Zhao, Jun Li, Yingjun Wang, Zuhua Guo, Xiaoyue Li (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/plos-pone-0352586.txt.
Description
Chenping Zhao and co-authors propose in 2026 a network for clearing haze from a single photo, where haze depth varies across the scene. Their design recalibrates features layer by layer, refines local detail through a coupled residual block, and gathers context at several scales through widened filters. A redesigned attention block uses convolutions instead of dense layers to model channel and space together.
Tests on standard benchmarks gave top scores on both reference based and blind quality measures, according to the authors. The system handled real world scenes with uneven haze well. Clearer images help later tasks such as driving vision and photo editing in poor weather.
Catalog record
| CREATOR | Chenping Zhao, Jun Li, Yingjun Wang, Zuhua Guo, Xiaoyue Li |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39153 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |