AI Paper Atlas

Research map / 3D vision / Depth and stereo

Long-range Turbulence Mitigation: A Large-scale Dataset and A Coarse-to-fine Framework

Shengqi Xu, Run Sun, Yi Chang, Shuning Cao, Xueyao Xiao, Luxin Yan

ECCV 2024 · 3D vision › Depth and stereo · approach: Non-line-of-sight imaging · 9 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 frequency-aware reference frame (FRF) registration (ablation)CDSP · Synthetic turbulencePSNR: 23.02 -> 23.96Improved
add subspace-based low-rank tensor refinement (SLRTR) (ablation)CDSP · Synthetic turbulencePSNR: 23.53 -> 23.96Improved
add both FRF and SLRTR over neither (ablation)CDSP · Synthetic turbulenceSSIM: 0.8051 -> 0.8385Improved
add SLRTR without FRF (ablation)CDSP · Synthetic turbulenceSSIM: 0.8051 -> 0.8034No clear effect
replace temporal-average reference frame with FRF in NDL (ablation)NDL · 5 Km synthetic turbulencePSNR: 22.95 -> 23.29Improved
replace temporal-average reference frame with FRF in TurbRecon (ablation)TurbRecon · 5 Km synthetic turbulencePSNR: 24.33 -> 24.79Improved
replace non-local average reference with FRF in CDSP (ablation)CDSP · 5 Km synthetic turbulencePSNR: 24.68 -> 25.41Improved
CDSP compared with TurbRecon as distance growsCDSP · Synthetic turbulencePSNR: 27.94 vs 27.27 (2 Km), 24.84 vs 23.95 (8 Km)Improved
apply CDSP restoration before text recognitionCRNN, ASTERN, DAN recognizers · TurbuTextrecognition accuracy: 0.2553 -> 0.5248 (CRNN)Improved
apply single-frame TurbNet restoration before CRNN recognitionCRNN · TurbuTextrecognition accuracy: 0.2553 -> 0.2057Hurt
remove subspace or self-similarity from SLRTR (ablation)SLRTR · qualitative on long-range turbulencevisual qualityHurt

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