Dezhen Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Dezhen Wang has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Yanlin He
Public Health Science and Engineering College, Tianjin University of Traditional Chinese Medicine, Tianjin, China
3 shared publications - Sheng Miao
School of Artificial Intelligence, Qingdao University of Technology, Qingdao 266520, China
2 shared publications - Wei Zhang
Harbin Institute of Technology
2 shared publications - Xiang Shen
The George Washington University Washington DC USA
2 shared publications - Zishun Song
Qingdao University of Technology
2 shared publications - Bin Jiang
China University of Petroleum (East China)
1 shared publication
Latest Publications
- 21 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 5404816
S 3 Mamba-Pan: Spectral–Spatial-Scale Mamba With Frequency-Decoupled Dual-Stream for Pansharpening
Pansharpening plays a critical role in remote sensing image processing by generating high-resolution multispectral (HRMS) images through the fusion of low-resolution multispectral (LRMS) and high-resolution panchromatic (PAN) images. In recent years, deep learning–based approaches have significantly advanced fusion performance via end-to-end modeling. However, existing methods still face notable limitations, including insufficient modeling of long-range dependencies, tight coupling between spectral and spatial information that may introduce spectral distortion and high computational complexity that restricts practical deployment in high-resolution scenarios. To address these issues, we propose Spectral-Spatial-Scale Mamba for Pansharpening (S3Mamba-Pan), a frequency-aware state space framework that jointly models spectral structure, spatial texture and multi-scale context in a coordinated yet decoupled manner. The method first extracts a global spectral anchor from the LRMS input to guide fusion, then performs wavelet-based frequency decoupling to separate low-frequency spectral components from high-frequency spatial details. A dual-stream Mamba architecture conducts parallel state space modeling for spectral correlation learning and PAN-guided texture enhancement and an adaptive distribution recalibration module aligns channel-wise statistics before reconstruction. Experiments on WV3, QuickBird and GF2 demonstrate consistent improvements over representative CNN, Transformer and SSM-based baselines under reduced-resolution evaluation. Under full-resolution no-reference assessment on WV3, S3Mamba-Pan achieves Dλ 0.0144, Ds 0.0303 and HQNR 0.960, indicating improved spectral fidelity and spatial consistency. Ablation and visualization analyses further corroborate the contribution of each component and provide interpretable evidence for the effectiveness of frequency decoupling and spectral anchoring. Code is available at https://github.com/FreeZS-a/S3Mamba.
- 21 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 5404816
S 3 Mamba-Pan: Spectral–Spatial-Scale Mamba With Frequency-Decoupled Dual-Stream for Pansharpening
Pansharpening plays a critical role in remote sensing image processing by generating high-resolution multispectral (HRMS) images through the fusion of low-resolution multispectral (LRMS) and high-resolution panchromatic (PAN) images. In recent years, deep learning–based approaches have significantly advanced fusion performance via end-to-end modeling. However, existing methods still face notable limitations, including insufficient modeling of long-range dependencies, tight coupling between spectral and spatial information that may introduce spectral distortion and high computational complexity that restricts practical deployment in high-resolution scenarios. To address these issues, we propose Spectral-Spatial-Scale Mamba for Pansharpening (S3Mamba-Pan), a frequency-aware state space framework that jointly models spectral structure, spatial texture and multi-scale context in a coordinated yet decoupled manner. The method first extracts a global spectral anchor from the LRMS input to guide fusion, then performs wavelet-based frequency decoupling to separate low-frequency spectral components from high-frequency spatial details. A dual-stream Mamba architecture conducts parallel state space modeling for spectral correlation learning and PAN-guided texture enhancement and an adaptive distribution recalibration module aligns channel-wise statistics before reconstruction. Experiments on WV3, QuickBird and GF2 demonstrate consistent improvements over representative CNN, Transformer and SSM-based baselines under reduced-resolution evaluation. Under full-resolution no-reference assessment on WV3, S3Mamba-Pan achieves Dλ 0.0144, Ds 0.0303 and HQNR 0.960, indicating improved spectral fidelity and spatial consistency. Ablation and visualization analyses further corroborate the contribution of each component and provide interpretable evidence for the effectiveness of frequency decoupling and spectral anchoring. Code is available at https://github.com/FreeZS-a/S3Mamba.
- Open Access
- 5 January 2026
- Published by Oxford University Press (OUP) in Journal of Computational Design and Engineering
- Vol. 13 (2) , 88-104
LMSGNet: Semantic prior and edge guided network for robust polyp segmentation
Automated polyp segmentation, which aims to accurately delineate polyp regions from colonoscopy images, is a critical task for computer-aided diagnosis in colorectal cancer prevention. Although many deep learning-based models have been proposed for this task, there are still some challenges. First, existing models still suffer performance degradation when confronted with small polyps, blurry boundaries, and cross-dataset testing, indicating limited robustness and generalization. Second, existing approaches predominantly focus on visual features, leaving the potential guidance of textual semantic information largely unexplored. To address these problems, we propose a novel Large-Model Semantics-Guided Network (LMSGNet) that leverages semantic guidance to achieve high-precision polyp segmentation. Specifically, we introduce semantic prompts encoded by Contrastive Language-Image Pre-Training (CLIP), employ a Multi-level Memory Router (MMR) to dynamically select relevant semantics, and incorporate a cross-modal attention (CMA) mechanism to enable bidirectional interactions between visual and semantic features, thereby enhancing global semantic consistency. In addition, we design a Semantic-Edge Guided Block (SEGB) combined with Multi-scale Edge Features (MSEF) to refine ambiguous boundaries and small targets, yielding synergistic global–local enhancement. Extensive experiments on five public datasets demonstrate that our model consistently outperforms existing state-of-the-art models. Two ablation studies respectively demonstrate the effectiveness of different modules and the contribution of semantic guidance.
- 5 January 2026
- Published by Oxford University Press (OUP) in Journal of Computational Design and Engineering
- Vol. 13 (2) , 88-104
Open AccessLMSGNet: Semantic prior and edge guided network for robust polyp segmentation
Automated polyp segmentation, which aims to accurately delineate polyp regions from colonoscopy images, is a critical task for computer-aided diagnosis in colorectal cancer prevention. Although many deep learning-based models have been proposed for this task, there are still some challenges. First, existing models still suffer performance degradation when confronted with small polyps, blurry boundaries, and cross-dataset testing, indicating limited robustness and generalization. Second, existing approaches predominantly focus on visual features, leaving the potential guidance of textual semantic information largely unexplored. To address these problems, we propose a novel Large-Model Semantics-Guided Network (LMSGNet) that leverages semantic guidance to achieve high-precision polyp segmentation. Specifically, we introduce semantic prompts encoded by Contrastive Language-Image Pre-Training (CLIP), employ a Multi-level Memory Router (MMR) to dynamically select relevant semantics, and incorporate a cross-modal attention (CMA) mechanism to enable bidirectional interactions between visual and semantic features, thereby enhancing global semantic consistency. In addition, we design a Semantic-Edge Guided Block (SEGB) combined with Multi-scale Edge Features (MSEF) to refine ambiguous boundaries and small targets, yielding synergistic global–local enhancement. Extensive experiments on five public datasets demonstrate that our model consistently outperforms existing state-of-the-art models. Two ablation studies respectively demonstrate the effectiveness of different modules and the contribution of semantic guidance.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- p. 1-23
DTF-Mamba: Dynamic Clustering and Topology-Aware Frequency-Enhanced Mamba Network for Remote Sensing Segmentation
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- p. 1-23
Open AccessDTF-Mamba: Dynamic Clustering and Topology-Aware Frequency-Enhanced Mamba Network for Remote Sensing Segmentation
- 18 June 2025
- Published by Elsevier in Computerized Medical Imaging and Graphics
- Vol. 124, 102580
Dual Stream Feature Fusion 3D Network for supraspinatus tendon tear classification
- 18 June 2025
- Published by Elsevier in Computerized Medical Imaging and Graphics
- Vol. 124, 102580
Dual Stream Feature Fusion 3D Network for supraspinatus tendon tear classification
- Open Access
- 16 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 154337-154348
Applying Hybrid Deep Learning Models to Assess Upper Limb Rehabilitation
Upper limb rehabilitation training is an effective method to restore and improve the upper limb motor function of stroke patients, which enables them to engage in daily life activities independently. However, traditional rehabilitation training methods have limitations such as high cost of rehabilitation equipment and imprecise assessment. Currently, Kinect is widely used to track key points of human skeleton, but Kinect is costly and requires high indoor environment, which makes it unsuitable for use in complex scenarios. In order to solve these problems, this study presents a new method that utilizes deep learning algorithms to recognize rehabilitation movements and thus evaluate rehabilitation effects. The proposed method is designed to help healthcare professionals monitor the patient’s rehabilitation progress and develop an optimal rehabilitation program for the patient. The method uses a monocular camera to capture video data from the patient’s upper limb rehabilitation training, utilizes Faster R-CNN and HRNet to recognize the human body position and upper limb bone key point information, and then builds a long short-term memory (LSTM) neural network model incorporating the ProbSparse Self-Attention mechanism to evaluate the rehabilitation training movements. The experimental results show that the accuracy of the proposed method in this study reaches 94.1%, and its overall performance is better than that of the baseline method in low-cost, high-precision and complex scenarios.
- 16 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 154337-154348
Open AccessApplying Hybrid Deep Learning Models to Assess Upper Limb Rehabilitation
Upper limb rehabilitation training is an effective method to restore and improve the upper limb motor function of stroke patients, which enables them to engage in daily life activities independently. However, traditional rehabilitation training methods have limitations such as high cost of rehabilitation equipment and imprecise assessment. Currently, Kinect is widely used to track key points of human skeleton, but Kinect is costly and requires high indoor environment, which makes it unsuitable for use in complex scenarios. In order to solve these problems, this study presents a new method that utilizes deep learning algorithms to recognize rehabilitation movements and thus evaluate rehabilitation effects. The proposed method is designed to help healthcare professionals monitor the patient’s rehabilitation progress and develop an optimal rehabilitation program for the patient. The method uses a monocular camera to capture video data from the patient’s upper limb rehabilitation training, utilizes Faster R-CNN and HRNet to recognize the human body position and upper limb bone key point information, and then builds a long short-term memory (LSTM) neural network model incorporating the ProbSparse Self-Attention mechanism to evaluate the rehabilitation training movements. The experimental results show that the accuracy of the proposed method in this study reaches 94.1%, and its overall performance is better than that of the baseline method in low-cost, high-precision and complex scenarios.
- 17 June 2023
- Published by Wiley in IEEJ Transactions on Electrical and Electronic Engineering
- Vol. 18 (8) , 1296-1302
Research on the Classification Method of Refrigerator Ingredients Based on Inception Structure
- 17 June 2023
- Published by Wiley in IEEJ Transactions on Electrical and Electronic Engineering
- Vol. 18 (8) , 1296-1302
Research on the Classification Method of Refrigerator Ingredients Based on Inception Structure
- 28 April 2000
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 87 (9) , 5188-5190
Pinhole imaging in magnetic tunnel junctions
Magnetic tunnel junctions may experience failure due to local short cuts in the insulating layer of such devices. The thicknesses of the insulating layers need to be reduced, which will likely make this problem more prevalent. To develop low resistance magnetic tunnel junctions, the density of pinholes must be analyzed. Using the electrodeposition of copper, we have developed a method for the imaging of pinholes. With the selective nucleation of copper at pinhole sites, structures are formed which can be visualized by conventional microscopy techniques. The potential applied for electrodeposition might cause dielectric breakdown of weak spots in the insulator layer. Variation of the applied voltage will allow identification of such sources of device failure.
- 28 April 2000
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 87 (9) , 5188-5190
Pinhole imaging in magnetic tunnel junctions
Magnetic tunnel junctions may experience failure due to local short cuts in the insulating layer of such devices. The thicknesses of the insulating layers need to be reduced, which will likely make this problem more prevalent. To develop low resistance magnetic tunnel junctions, the density of pinholes must be analyzed. Using the electrodeposition of copper, we have developed a method for the imaging of pinholes. With the selective nucleation of copper at pinhole sites, structures are formed which can be visualized by conventional microscopy techniques. The potential applied for electrodeposition might cause dielectric breakdown of weak spots in the insulator layer. Variation of the applied voltage will allow identification of such sources of device failure.