PGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusion

Abstract High Dynamic Range (HDR) imaging captures the complete luminance information of a scene by fusing a multi-exposure image sequence. However, a key challenge remains: ghosting artifacts caused by object motion in dynamic scenes. Existing methods typically adopt an align-then-fuse strategy, of...

Full description

Saved in:
Bibliographic Details
Main Authors: Ying Qi, Qiushi Li, Zhaoyuan Huang, Jian Li, Chenyang Wang, Teng Wan, Qiang Zhang
Format: Article
Language:English
Published: Elsevier 2025-08-01
Series:Journal of King Saud University: Computer and Information Sciences
Subjects:
Online Access:https://doi.org/10.1007/s44443-025-00230-z
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849225881321472000
author Ying Qi
Qiushi Li
Zhaoyuan Huang
Jian Li
Chenyang Wang
Teng Wan
Qiang Zhang
author_facet Ying Qi
Qiushi Li
Zhaoyuan Huang
Jian Li
Chenyang Wang
Teng Wan
Qiang Zhang
author_sort Ying Qi
collection DOAJ
description Abstract High Dynamic Range (HDR) imaging captures the complete luminance information of a scene by fusing a multi-exposure image sequence. However, a key challenge remains: ghosting artifacts caused by object motion in dynamic scenes. Existing methods typically adopt an align-then-fuse strategy, often overlooking the spatial variability of alignment quality, which makes it difficult to balance ghosting suppression and detail preservation when handling complex motion. To address this issue, this paper proposes PGHDR, which leverages a quality-aware progressive alignment strategy. The core innovation is a learning paradigm that jointly optimizes motion estimation and alignment quality perception. This approach treats alignment quality as a learnable, spatially-varying confidence and materializes this concept through three synergistic modules. The Progressive Deformable Feature Alignment (PDFA) module achieves robust feature extraction through a two-stage deformable convolution and exposure-aware modulation. The Quality-Guided Motion Compensation (QGMC) module jointly learns optical flow and a confidence mask, enabling reliability-aware motion estimation. The Structure-Preserving Adaptive Fusion (SPAF) module performs weighted fusion based on the learned quality information while preserving image details via a local structure constraint. Experiments show that PGHDR achieves a PSNR- $$\ell $$ ℓ of 42.58 dB and an HDR-VDP-2 score of 67.15. Visual evaluations on multiple datasets demonstrate that our method significantly outperforms existing approaches, achieving high-quality, ghost-free HDR reconstruction.
format Article
id doaj-art-f08df7a357b146e2bea2beed57faf88f
institution Kabale University
issn 1319-1578
2213-1248
language English
publishDate 2025-08-01
publisher Elsevier
record_format Article
series Journal of King Saud University: Computer and Information Sciences
spelling doaj-art-f08df7a357b146e2bea2beed57faf88f2025-08-24T11:53:37ZengElsevierJournal of King Saud University: Computer and Information Sciences1319-15782213-12482025-08-0137712310.1007/s44443-025-00230-zPGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusionYing Qi0Qiushi Li1Zhaoyuan Huang2Jian Li3Chenyang Wang4Teng Wan5Qiang Zhang6School of Computer Science and Engineering, Northwest Normal UniversitySchool of Computer Science and Engineering, Northwest Normal UniversitySchool of Computer Science and Engineering, Northwest Normal UniversitySchool of Computer Science and Engineering, Northwest Normal UniversitySchool of Computer Science and Engineering, Northwest Normal UniversitySchool of Computer Science and Engineering, Northwest Normal UniversitySchool of Computer Science and Engineering, Northwest Normal UniversityAbstract High Dynamic Range (HDR) imaging captures the complete luminance information of a scene by fusing a multi-exposure image sequence. However, a key challenge remains: ghosting artifacts caused by object motion in dynamic scenes. Existing methods typically adopt an align-then-fuse strategy, often overlooking the spatial variability of alignment quality, which makes it difficult to balance ghosting suppression and detail preservation when handling complex motion. To address this issue, this paper proposes PGHDR, which leverages a quality-aware progressive alignment strategy. The core innovation is a learning paradigm that jointly optimizes motion estimation and alignment quality perception. This approach treats alignment quality as a learnable, spatially-varying confidence and materializes this concept through three synergistic modules. The Progressive Deformable Feature Alignment (PDFA) module achieves robust feature extraction through a two-stage deformable convolution and exposure-aware modulation. The Quality-Guided Motion Compensation (QGMC) module jointly learns optical flow and a confidence mask, enabling reliability-aware motion estimation. The Structure-Preserving Adaptive Fusion (SPAF) module performs weighted fusion based on the learned quality information while preserving image details via a local structure constraint. Experiments show that PGHDR achieves a PSNR- $$\ell $$ ℓ of 42.58 dB and an HDR-VDP-2 score of 67.15. Visual evaluations on multiple datasets demonstrate that our method significantly outperforms existing approaches, achieving high-quality, ghost-free HDR reconstruction.https://doi.org/10.1007/s44443-025-00230-zHigh dynamic range (HDR) imagingQuality-aware fusionDeformable convolutionMotion estimationDeep learning
spellingShingle Ying Qi
Qiushi Li
Zhaoyuan Huang
Jian Li
Chenyang Wang
Teng Wan
Qiang Zhang
PGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusion
Journal of King Saud University: Computer and Information Sciences
High dynamic range (HDR) imaging
Quality-aware fusion
Deformable convolution
Motion estimation
Deep learning
title PGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusion
title_full PGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusion
title_fullStr PGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusion
title_full_unstemmed PGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusion
title_short PGHDR: Dynamic HDR reconstruction with progressive feature alignment and quality-guided fusion
title_sort pghdr dynamic hdr reconstruction with progressive feature alignment and quality guided fusion
topic High dynamic range (HDR) imaging
Quality-aware fusion
Deformable convolution
Motion estimation
Deep learning
url https://doi.org/10.1007/s44443-025-00230-z
work_keys_str_mv AT yingqi pghdrdynamichdrreconstructionwithprogressivefeaturealignmentandqualityguidedfusion
AT qiushili pghdrdynamichdrreconstructionwithprogressivefeaturealignmentandqualityguidedfusion
AT zhaoyuanhuang pghdrdynamichdrreconstructionwithprogressivefeaturealignmentandqualityguidedfusion
AT jianli pghdrdynamichdrreconstructionwithprogressivefeaturealignmentandqualityguidedfusion
AT chenyangwang pghdrdynamichdrreconstructionwithprogressivefeaturealignmentandqualityguidedfusion
AT tengwan pghdrdynamichdrreconstructionwithprogressivefeaturealignmentandqualityguidedfusion
AT qiangzhang pghdrdynamichdrreconstructionwithprogressivefeaturealignmentandqualityguidedfusion