AI Paper Atlas

Research map / Image generation / Image restoration

UMNet: Uncertainty-guided Memory Network for Hyperspectral Pansharpening

Xiaozheng Wang, Yong Yang, Shuying Huang, Nayu Liu, Ziyang Liu

AAAI 2026 · Image generation › Image restoration · approach: Pansharpening and spectral fusion · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add SRFU (with sequence modeling) instead of LFMamba state-space model (ablation)UMNet · Pavia Center (simulated)PSNR: 37.8800 -> 39.0832Improved
add sequence modeling to SRFU (ablation)UMNet · Pavia Center (simulated)PSNR: 38.7148 -> 39.0832Improved
use SMIU (NMF) instead of simple addition (ablation)UMNet · Pavia Center (simulated)PSNR: 38.7210 -> 39.0832Improved
use uncertainty-based losses (LEU) instead of L1 (ablation)UMNet · Pavia Center (simulated)PSNR: 38.7714 -> 38.8343Improved
add uncertainty-guided reconstruction loss LUG to LEU (ablation)UMNet · Pavia Center (simulated)PSNR: 38.8343 -> 39.0832Improved
use both spatial and channel uncertainty instead of one (ablation)UMNet · Pavia Center (simulated)PSNR: 38.9200 / 38.9117 -> 39.0832Improved
use 8 channels per group instead of 4 or 16 (ablation)UMNet · Pavia Center (simulated)PSNR: 37.4640 (4), 38.0425 (16) -> 39.0832 (8)Improved
set SMIU latent dimension r=c/4 instead of c/2 or c/8 (ablation)UMNet · Pavia Center (simulated)PSNR: 38.4235 (c/2), 38.4771 (c/8) -> 39.0832Improved
use UMNet instead of best baselineHS pansharpening · Chikusei (simulated)PSNR: 43.4759 -> 43.9936Improved
use UMNet instead of best baseline on real dataHS pansharpening · FR1 (real)QNR: 0.9576 -> 0.9606Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy