Shun Lv
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Shun Lv has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
- Yiguang Liu
Sichuan University
14 shared publications - Tianhang Tang
Sichuan University
5 shared publications - Jie Chen
Sichuan University
4 shared publications - Zeyu Chen
College of Computer Science, Sichuan University, Chengdu, China
4 shared publications - Shaobing Gao
Sichuan University
3 shared publications Zhicheng Yu
3 shared publications
Latest Publications
- 16 November 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (23) , 6423
Competition-aware Poisson structural branch probability fields for single-photon imaging
Single-photon depth and intensity reconstruction is challenging at low signal-to-background ratios, where background photons and Poisson fluctuations can create competing temporal branches. Spatially nonuniform backgrounds further make local branch evidence uneven, increasing the risk of branch misselection. We propose competition-aware Poisson structural branch probability fields (Poisson-SBPF) for support-adaptive branch stabilization in low-SBR reconstruction. Poisson-SBPF compares the apparent dominant branch with the strongest temporally separated competitor and uses their support-normalized Poisson evidence gap to measure branch separability. Compact support is retained for separable decisions, whereas enlarged support is invoked only for fragile branch competition. Depth and intensity are estimated from the same stabilized branch. On simulated and real-response data, Poisson-SBPF reduces average depth RMSE by about 35% and 16% relative to UA, respectively, and lowers the depth outlier fraction by about 76% in both cases.
- 16 November 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (23) , 6423
Competition-aware Poisson structural branch probability fields for single-photon imaging
Single-photon depth and intensity reconstruction is challenging at low signal-to-background ratios, where background photons and Poisson fluctuations can create competing temporal branches. Spatially nonuniform backgrounds further make local branch evidence uneven, increasing the risk of branch misselection. We propose competition-aware Poisson structural branch probability fields (Poisson-SBPF) for support-adaptive branch stabilization in low-SBR reconstruction. Poisson-SBPF compares the apparent dominant branch with the strongest temporally separated competitor and uses their support-normalized Poisson evidence gap to measure branch separability. Compact support is retained for separable decisions, whereas enlarged support is invoked only for fragile branch competition. Depth and intensity are estimated from the same stabilized branch. On simulated and real-response data, Poisson-SBPF reduces average depth RMSE by about 35% and 16% relative to UA, respectively, and lowers the depth outlier fraction by about 76% in both cases.
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Microwave Theory and Techniques
- Vol. 74 (11) , 9502-9518
Physical-Statistical Information Network for Unsupervised Sparse Reconstruction in Near-Field Millimeter-Wave SAR
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Microwave Theory and Techniques
- Vol. 74 (11) , 9502-9518
Physical-Statistical Information Network for Unsupervised Sparse Reconstruction in Near-Field Millimeter-Wave SAR
- Open Access
- 23 July 2026
- Published by Elsevier in SSRN Electronic Journal
Rethinking Monocular Depth Embedding for Generalized Stereo Matching
Generally, monocular methods capture rich contextual priors but lack geometric precision, whereas stereo methods are geometrically accurate yet struggle in textureless and occluded regions. Several approaches attempt to combine their strengths to enhance the generalization of stereo matching (SM) by aligning monocular depth with stereo information. However, establishing a stable and generalizable alignment is challenging, and unreliable monocular cues can substantially degrade performance. This paper rethinks monocular depth embedding. First, to prevent shortcut learning, we reduce branch coupling instead of expanding network width. Second, we construct soft constraints instead of hard ones from monocular depth to improve tolerance to monocular depth errors. Based on the principles, we integrate monocular information into both feature extraction and GRU iterations. Specifically, the monocular depth map is fused with the RGB image to sharpen depth boundary perception and suppress matching ambiguities. The fused image is then used for feature extraction, allowing the contextual features to encode global geometric information. Furthermore, the monocular depth gradient feature is employed to guide disparity updates, helping to escape local oscillations. Finally, to address the boundary blurring of supervised disparity caused by data augmentation, we propose an edge confidence estimation method and an edge-aware loss function. Our method achieves state-of-the-art (SOTA) performance on multiple standard benchmarks, demonstrating excellent generalization while improving accuracy.
- 23 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessRethinking Monocular Depth Embedding for Generalized Stereo Matching
Generally, monocular methods capture rich contextual priors but lack geometric precision, whereas stereo methods are geometrically accurate yet struggle in textureless and occluded regions. Several approaches attempt to combine their strengths to enhance the generalization of stereo matching (SM) by aligning monocular depth with stereo information. However, establishing a stable and generalizable alignment is challenging, and unreliable monocular cues can substantially degrade performance. This paper rethinks monocular depth embedding. First, to prevent shortcut learning, we reduce branch coupling instead of expanding network width. Second, we construct soft constraints instead of hard ones from monocular depth to improve tolerance to monocular depth errors. Based on the principles, we integrate monocular information into both feature extraction and GRU iterations. Specifically, the monocular depth map is fused with the RGB image to sharpen depth boundary perception and suppress matching ambiguities. The fused image is then used for feature extraction, allowing the contextual features to encode global geometric information. Furthermore, the monocular depth gradient feature is employed to guide disparity updates, helping to escape local oscillations. Finally, to address the boundary blurring of supervised disparity caused by data augmentation, we propose an edge confidence estimation method and an edge-aware loss function. Our method achieves state-of-the-art (SOTA) performance on multiple standard benchmarks, demonstrating excellent generalization while improving accuracy.
- 3 June 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (12) , 3317-3320
Fourier single-pixel imaging for dynamic color scene reconstruction via joint priors
Single-pixel imaging enables high-sensitivity measurements across diverse spectral regimes. However, reconstructing high-quality dynamic color scenes under sub-Nyquist sampling remains challenging due to the trade-off between spatiotemporal resolution and reconstruction fidelity. This article proposes a dynamic color Fourier single-pixel imaging framework that integrates a multi-channel acquisition scheme with a joint-prior reconstruction model to address this challenge. A hyper-Laplacian prior is employed to exploit cross-channel correlations, while a local low-rank prior captures temporal redundancy in dynamic scenes. Compared to a single prior, the joint approach enables mutual reinforcement between the two priors. Based on this formulation, we derive an efficient alternating minimization scheme that achieves significantly accelerated convergence through closed-form solutions. Experimental results demonstrate that the proposed method yields a PSNR improvement of over 5 dB compared to conventional Fourier single-pixel imaging, while maintaining robust reconstruction within 92 ms per frame.
- 3 June 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (12) , 3317-3320
Fourier single-pixel imaging for dynamic color scene reconstruction via joint priors
Single-pixel imaging enables high-sensitivity measurements across diverse spectral regimes. However, reconstructing high-quality dynamic color scenes under sub-Nyquist sampling remains challenging due to the trade-off between spatiotemporal resolution and reconstruction fidelity. This article proposes a dynamic color Fourier single-pixel imaging framework that integrates a multi-channel acquisition scheme with a joint-prior reconstruction model to address this challenge. A hyper-Laplacian prior is employed to exploit cross-channel correlations, while a local low-rank prior captures temporal redundancy in dynamic scenes. Compared to a single prior, the joint approach enables mutual reinforcement between the two priors. Based on this formulation, we derive an efficient alternating minimization scheme that achieves significantly accelerated convergence through closed-form solutions. Experimental results demonstrate that the proposed method yields a PSNR improvement of over 5 dB compared to conventional Fourier single-pixel imaging, while maintaining robust reconstruction within 92 ms per frame.
- 22 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 5623815
EntSeg: Entropy-Guided Pseudolabel Denoising and Masked Image Consistency for Cross-Domain Remote Sensing Segmentation
Unsupervised Domain Adaptation (UDA) for semantic segmentation of Remote Sensing Images (RSIs) aims to adapt a model trained on labeled source-domain data to a target domain without requiring target-domain annotations. Existing self-training (ST) approaches adapt the model by generating pseudo labels for target-domain samples. However, due to the inherent domain gap, these pseudo labels often contain substantial noise, which in turn limits the effectiveness of such methods. We observe that noisy pseudo labels typically exhibit higher entropy. Motivated by this observation, we propose entropy-guided segmentation (EntSeg) framework, a novel entropy-guided framework for cross-domain RSIs semantic segmentation, that comprising two key models. The entropy-weighted pseudo-label denoising (EPLD) module dynamically adjusts the learning weights of noisy pixel locations (i.e., pixels with high entropy) to mitigate the adverse influence of noisy pseudo labels on domain adaptation. Nevertheless, the dynamic weighting mechanism of EPLD also results in insufficient contextual learning for target-domain data, especially at locations assigned low weights. To address this issue, we introduce entropy-guided masked image consistency (EMIC), which employs an entropy-guided strategy to mask high-entropy regions and learns spatial contextual relationships in the target domain through masked image consistency, thereby enhancing the model’s robustness in visual recognition. Experimental results demonstrate that the proposed method achieves highly competitive performance. The code will be released at https://github.com/woldier/EntSeg.
- 22 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 5623815
EntSeg: Entropy-Guided Pseudolabel Denoising and Masked Image Consistency for Cross-Domain Remote Sensing Segmentation
Unsupervised Domain Adaptation (UDA) for semantic segmentation of Remote Sensing Images (RSIs) aims to adapt a model trained on labeled source-domain data to a target domain without requiring target-domain annotations. Existing self-training (ST) approaches adapt the model by generating pseudo labels for target-domain samples. However, due to the inherent domain gap, these pseudo labels often contain substantial noise, which in turn limits the effectiveness of such methods. We observe that noisy pseudo labels typically exhibit higher entropy. Motivated by this observation, we propose entropy-guided segmentation (EntSeg) framework, a novel entropy-guided framework for cross-domain RSIs semantic segmentation, that comprising two key models. The entropy-weighted pseudo-label denoising (EPLD) module dynamically adjusts the learning weights of noisy pixel locations (i.e., pixels with high entropy) to mitigate the adverse influence of noisy pseudo labels on domain adaptation. Nevertheless, the dynamic weighting mechanism of EPLD also results in insufficient contextual learning for target-domain data, especially at locations assigned low weights. To address this issue, we introduce entropy-guided masked image consistency (EMIC), which employs an entropy-guided strategy to mask high-entropy regions and learns spatial contextual relationships in the target domain through masked image consistency, thereby enhancing the model’s robustness in visual recognition. Experimental results demonstrate that the proposed method achieves highly competitive performance. The code will be released at https://github.com/woldier/EntSeg.
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35 (10577149) , 4832-4846
Infrared and Visible Image Fusion Using Bimodal Neuron and Dynamic Receptive Field Mechanisms
Infrared and visible image fusion (IVIF) significantly enhances scene interpretation by integrating broad-spectrum information. Drawing inspiration from specific snakes that possess an evolutionarily optimized bimodal sensory system capable of parallel processing infrared and visible radiation, we propose a novel IVIF framework incorporating two key elements: nonlinear cross-modal interactions across six distinct classes of snake bimodal neurons and dynamic center-surround receptive field organization. These biological principles are mathematically formalized and integrated within a deep neural network (DNN), optimized through an object detection region-guided loss and a frequency-dependent fusion loss that enable data-driven fusion strategy learning. Experimental results demonstrate that the optimized model effectively emulates the infrared-visible information integration observed in snake bimodal neurons. Critically, the nonlinear bimodal neurons capture a significantly greater amount of edge information and finer mid-to-high-frequency details, which are essential for the subsequent reconstruction of the fused image. Furthermore, a comprehensive evaluation of visual quality, encompassing both qualitative and quantitative assessments on six datasets, along with extensive object detection and semantic segmentation experiments using the fused images in both daytime and nighttime scenarios, demonstrates that our model outperforms traditional biologically-inspired IVIF algorithms, achieving performance comparable to SOTA DNN-based methods.
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 35 (10577149) , 4832-4846
Infrared and Visible Image Fusion Using Bimodal Neuron and Dynamic Receptive Field Mechanisms
Infrared and visible image fusion (IVIF) significantly enhances scene interpretation by integrating broad-spectrum information. Drawing inspiration from specific snakes that possess an evolutionarily optimized bimodal sensory system capable of parallel processing infrared and visible radiation, we propose a novel IVIF framework incorporating two key elements: nonlinear cross-modal interactions across six distinct classes of snake bimodal neurons and dynamic center-surround receptive field organization. These biological principles are mathematically formalized and integrated within a deep neural network (DNN), optimized through an object detection region-guided loss and a frequency-dependent fusion loss that enable data-driven fusion strategy learning. Experimental results demonstrate that the optimized model effectively emulates the infrared-visible information integration observed in snake bimodal neurons. Critically, the nonlinear bimodal neurons capture a significantly greater amount of edge information and finer mid-to-high-frequency details, which are essential for the subsequent reconstruction of the fused image. Furthermore, a comprehensive evaluation of visual quality, encompassing both qualitative and quantitative assessments on six datasets, along with extensive object detection and semantic segmentation experiments using the fused images in both daytime and nighttime scenarios, demonstrates that our model outperforms traditional biologically-inspired IVIF algorithms, achieving performance comparable to SOTA DNN-based methods.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 12, 1814-1827
Fisher–Rao Adaptive Reconstruction for Single-Photon LiDAR in Challenging Environments
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 12, 1814-1827
Fisher–Rao Adaptive Reconstruction for Single-Photon LiDAR in Challenging Environments
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 12, 1392-1406
Single-Photon Neural Assumed-Density Filter for Dynamic Scene Reconstruction
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 12, 1392-1406
Single-Photon Neural Assumed-Density Filter for Dynamic Scene Reconstruction
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 33, 3182-3186
Multi-Scale K-Cosine Curvature Fusion With Nonlinear Measure for Robust Corner Detection
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 33, 3182-3186
Multi-Scale K-Cosine Curvature Fusion With Nonlinear Measure for Robust Corner Detection