DEMANet: A dehazing enhanced multi-branch attention network for remote sensing images

Peixue Liu, Shu Liu, Pengfei He, Guoheng Wang · 2026

Read DEMANet: A dehazing enhanced multi-branch attention network for remote sensing images by Peixue Liu, Shu Liu, Pengfei He, Guoheng Wang (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-0354121.txt.

Description

Peixue Liu and coauthors published a 2026 computer vision paper in PLOS ONE on clearing haze from satellite images. Haze dulls the images that land-use mapping and change detection depend on, washing out detail and color. Their network, DEMANet, processes the image through three shrinking stages: early layers use a dual-path module to protect fine detail. Deep layers use two-level attention to grasp the whole scene and haze pattern. A cross-stage module uses edges and color differences to locate haze and ferry detail between stages.

On widely used public remote sensing datasets, the network beat existing dehazing algorithms while keeping intricate detail and faithful color. The authors attribute the gains to the division of labor: local branches guard texture, attention guards global consistency, and the haze-guided links stop information leaking away during reconstruction. Cleaner satellite imagery feeds directly into better environmental monitoring and land management.

Catalog record

CREATORPeixue Liu, Shu Liu, Pengfei He, Guoheng Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39156
LICENSECC_BY
RIGHTSCC BY
SOURCEPLOS

Source: https://doi.org/10.1371/journal.pone.0354121

Browse the full library · About the Tower of Babel