Zhuoran Zheng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zhuoran Zheng has a total of 106 co-authors. Here are the top co-authors by number of shared publications:
- Dianjie Lu
Shandong Normal University, Jinan, China
11 shared publications - Chen Wu
National university of defense technology, Changsha, China
10 shared publications - Guijuan Zhang
Shandong Normal University, Jinan, China
10 shared publications Chen Wu
8 shared publications- Chen Lyu
Shandong Normal University, Jinan, China
6 shared publications - Pu Wang
School of Mathematics, Shandong University, Jinan, China
6 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 2998-3006
UHD-GPGNet: UHD Video Denoising via Gaussian-Process-Guided Local Spatio-Temporal Modeling
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 2998-3006
Open AccessUHD-GPGNet: UHD Video Denoising via Gaussian-Process-Guided Local Spatio-Temporal Modeling
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 10701-10709
Clifford-Algebra-Based Geometric Feature Fusion for Real-Time UHD Low-Light Image Enhancement
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 10701-10709
Open AccessClifford-Algebra-Based Geometric Feature Fusion for Real-Time UHD Low-Light Image Enhancement
- 14 September 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 334
Focusing on the essentials: Reliability-driven sparse representation for infrared small target detection
- 14 September 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 334
Focusing on the essentials: Reliability-driven sparse representation for infrared small target detection
- Open Access
- 17 April 2026
- Published by Elsevier in SSRN Electronic Journal
Hazy Pedestrian Trajectory Prediction via Physical Priors and Graph-Mamba
Pedestrian trajectory prediction is fundamental to intelligent transportation systems, yet existing methods assume clean inputs and ignore the cascade degradation in hazy environments, where atmospheric scattering degrades images, impairs detection and tracking, and produces noisy trajectory observations that propagate errors into prediction. To address this problem, we propose FAPTP, a physics-informed framework comprising three tightly coupled modules. PhyFusion constructs a di erentiable atmospheric scattering model grounded in Koschmieder theory, producing haze-invariant features via physics-constrained reconstruction. MambaST-Sequences introduces a selective state-space model at O(T) complexity to lter noise-corrupted temporal signals while preserving long-range motion dependencies. DynaHeteroNet develops a visibility-aware heterogeneous graph network that conditions multi-granularity interaction weights on haze density and detection con dence with scene-adaptive mechanisms. We further construct a detectioncascaded hazy benchmark on ETH/UCY with behavioral adaptation across four haze levels (β ∈ {0,0.5,1.0,2.0}), where all methods receive trajectories from a shared YOLOv8+ByteTrack pipeline. Extensive experiments demonstrate that FAPTP consistently achieves the best or co-best performance
- 17 April 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessHazy Pedestrian Trajectory Prediction via Physical Priors and Graph-Mamba
Pedestrian trajectory prediction is fundamental to intelligent transportation systems, yet existing methods assume clean inputs and ignore the cascade degradation in hazy environments, where atmospheric scattering degrades images, impairs detection and tracking, and produces noisy trajectory observations that propagate errors into prediction. To address this problem, we propose FAPTP, a physics-informed framework comprising three tightly coupled modules. PhyFusion constructs a di erentiable atmospheric scattering model grounded in Koschmieder theory, producing haze-invariant features via physics-constrained reconstruction. MambaST-Sequences introduces a selective state-space model at O(T) complexity to lter noise-corrupted temporal signals while preserving long-range motion dependencies. DynaHeteroNet develops a visibility-aware heterogeneous graph network that conditions multi-granularity interaction weights on haze density and detection con dence with scene-adaptive mechanisms. We further construct a detectioncascaded hazy benchmark on ETH/UCY with behavioral adaptation across four haze levels (β ∈ {0,0.5,1.0,2.0}), where all methods receive trajectories from a shared YOLOv8+ByteTrack pipeline. Extensive experiments demonstrate that FAPTP consistently achieves the best or co-best performance
- 10 April 2026
- Published by Elsevier in Neurocomputing
- Vol. 685
UHD image dehazing via anDehazeFormer with atmospheric-aware KV cache
- 10 April 2026
- Published by Elsevier in Neurocomputing
- Vol. 685
UHD image dehazing via anDehazeFormer with atmospheric-aware KV cache
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 1-15
FS-FlowNet: Frequency–Spatial Dual-Domain Residual Flow Network for Remote Sensing Dehazing
Atmospheric haze and cloud cover severely degrade remote sensing image quality, adversely affecting critical tasks like object recognition and change detection. Existing dehazing methods face significant challenges with such images: traditional physical models often cause over-dehazing and color distortion, while single-stage deep learning models generalize poorly due to fixed regression paths. To address these limitations, this paper proposes FS-FlowNet, a dual-domain frequency-spatial residual flow network for remote sensing image dehazing. We innovatively combine frequency-spatial domain processing with multi-stage ordinary differential equation optimization within an efficient end-to-end architecture. First, during coarse dehazing, an enhanced U-Net encoder–decoder conducts multi-layer feature extraction with skip connections to capture haze-related features while preserving spatial structures. Next, a dual-domain module extracts amplitude and phase spectra via FFT, using a dual-branch (frequency-spatial) strategy to improve high-frequency detail. We employ a flow matching algorithm to cast dehazing as a multi-stage flow-field optimization problem; Here, a fourth-order Runge–Kutta (RK4) solver is used to activate/solve the underlying ordinary differential equations, yielding smooth and physically plausible flow transformations. Extensive experiments on the real world benchmark and multiple synthetic datasets demonstrate superior performance in both real and synthetic scenarios.
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 1-15
FS-FlowNet: Frequency–Spatial Dual-Domain Residual Flow Network for Remote Sensing Dehazing
Atmospheric haze and cloud cover severely degrade remote sensing image quality, adversely affecting critical tasks like object recognition and change detection. Existing dehazing methods face significant challenges with such images: traditional physical models often cause over-dehazing and color distortion, while single-stage deep learning models generalize poorly due to fixed regression paths. To address these limitations, this paper proposes FS-FlowNet, a dual-domain frequency-spatial residual flow network for remote sensing image dehazing. We innovatively combine frequency-spatial domain processing with multi-stage ordinary differential equation optimization within an efficient end-to-end architecture. First, during coarse dehazing, an enhanced U-Net encoder–decoder conducts multi-layer feature extraction with skip connections to capture haze-related features while preserving spatial structures. Next, a dual-domain module extracts amplitude and phase spectra via FFT, using a dual-branch (frequency-spatial) strategy to improve high-frequency detail. We employ a flow matching algorithm to cast dehazing as a multi-stage flow-field optimization problem; Here, a fourth-order Runge–Kutta (RK4) solver is used to activate/solve the underlying ordinary differential equations, yielding smooth and physically plausible flow transformations. Extensive experiments on the real world benchmark and multiple synthetic datasets demonstrate superior performance in both real and synthetic scenarios.
- 23 January 2026
- Published by Springer Nature
- p. 121-135
ChebSpec-Net: Linear Spectral Graph Restoration for UHD Images
- 23 January 2026
- Published by Springer Nature
- p. 121-135
ChebSpec-Net: Linear Spectral Graph Restoration for UHD Images
- 22 January 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 312
Fusion requires interaction: a hybrid Mamba-transformer architecture for deep interactive fusion of multi-modal images
- 22 January 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 312
Fusion requires interaction: a hybrid Mamba-transformer architecture for deep interactive fusion of multi-modal images