Shuying Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Shuying Li has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Chuang Yang
The Hong Kong Polytechnic University
6 shared publications San Zhang
School of Artificial Intelligence and the School of Automation, Xian University of Posts and Telecommunications, Xian, China
5 shared publications- Qiang Li
School of Artificial Intelligence, OPtics and ElectroNics (iOPEN), Northwestern Polytechnical University, Xian, China
3 shared publications Qiang Ma
School of Automation, Xian University of Posts and Telecommunications, Xian, China
2 shared publications- Wuwei Wang
School of Automation, Xi’an University of Posts and Telecommunications, Xi’an, China
2 shared publications Yuchen Wang
School of Artificial Intelligence and the School of Automation, Xian University of Posts and Telecommunications, Xian, China
2 shared publications
Latest Publications
- 5 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 5625017
CTTF: Cross-Temporal Tri-Difference Fusion With Mamba for Remote Sensing Change Detection
Mamba has recently been widely applied to remote sensing change detection tasks due to its ability to efficiently represent long-range dependencies with nearly linear computational complexity. However, existing Mamba-based change detection methods focus primarily on global modeling during the decoding stage, where insufficient attention to fine-grained details causes blurred change boundaries. In addition, most existing methods rely solely on a single type of differential feature (e.g., signed difference or absolute difference), ignoring the complementary information among different types of differential features, which makes them prone to missed or false detections under registration errors. To address these issues, this paper proposes a decoder-side improved framework named CTTF (Cross-Temporal Tri-Diff Fusion). It consists of a GLP (Global–Local–Point) Block and a SASF (Signed–Absolute–Sum Fusion) Module, enabling accurate boundary feature detection, thereby achieving more consistent and reliable change discrimination. Specifically, the GLP Block decouples and models boundary details, global semantics, and information anchors, achieving a dynamic balance between local enhancement and semantic fidelity, thereby significantly improving the boundary feature perception. The SASF constructs a tri-differential mechanism, where the sum difference serves as a stability prior to perform multi-scale calibration between the signed difference and the absolute difference. This design compensates for the loss of fine structural and consistency information caused by relying on a single differential domain, effectively reducing both missed and false detections. The proposed method achieves state-of-the-art (SOTA) performance on the WHU-CD, LEVIR-CD, SYSU-CD and CLCD datasets. Ablation experiments further validate the effectiveness of the proposed SASF Module and GLP Block.
- 5 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 5625017
CTTF: Cross-Temporal Tri-Difference Fusion With Mamba for Remote Sensing Change Detection
Mamba has recently been widely applied to remote sensing change detection tasks due to its ability to efficiently represent long-range dependencies with nearly linear computational complexity. However, existing Mamba-based change detection methods focus primarily on global modeling during the decoding stage, where insufficient attention to fine-grained details causes blurred change boundaries. In addition, most existing methods rely solely on a single type of differential feature (e.g., signed difference or absolute difference), ignoring the complementary information among different types of differential features, which makes them prone to missed or false detections under registration errors. To address these issues, this paper proposes a decoder-side improved framework named CTTF (Cross-Temporal Tri-Diff Fusion). It consists of a GLP (Global–Local–Point) Block and a SASF (Signed–Absolute–Sum Fusion) Module, enabling accurate boundary feature detection, thereby achieving more consistent and reliable change discrimination. Specifically, the GLP Block decouples and models boundary details, global semantics, and information anchors, achieving a dynamic balance between local enhancement and semantic fidelity, thereby significantly improving the boundary feature perception. The SASF constructs a tri-differential mechanism, where the sum difference serves as a stability prior to perform multi-scale calibration between the signed difference and the absolute difference. This design compensates for the loss of fine structural and consistency information caused by relying on a single differential domain, effectively reducing both missed and false detections. The proposed method achieves state-of-the-art (SOTA) performance on the WHU-CD, LEVIR-CD, SYSU-CD and CLCD datasets. Ablation experiments further validate the effectiveness of the proposed SASF Module and GLP Block.
- Top Cited Papers
- 5 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 1-13
HA2F: Dual-Module Collaboration-Guided Hierarchical Adaptive Aggregation Framework for Remote Sensing Change Detection
Remote sensing change detection (RSCD) aims to identify the spatio-temporal changes of land cover, providing critical support for multi-disciplinary applications (e.g., environmental monitoring, disaster assessment, and climate change studies). Existing methods focus either on extracting features from localized patches, or pursue processing entire images holistically, which leads to the cross temporal feature matching deviation and exhibiting sensitivity to radiometric and geometric noise. Following the above issues, we propose a dual-module collaboration guided hierarchical adaptive aggregation framework, namely HA2F, which consists of dynamic hierarchical feature calibration module (DHFCM) and noise-adaptive feature refinement module (NAFRM). The former dynamically fuses adjacent-level features through perceptual feature selection, suppressing irrelevant discrepancies to address multi-temporal feature alignment deviations. The NAFRM utilizes the dual feature selection mechanism to highlight the change sensitive regions and generate spatial masks, suppressing the interference of irrelevant regions or shadows. Extensive experiments verify the effectiveness of the proposed HA2F, which achieves state-of-the-art performance on LEVIR-CD, WHU-CD, and SYSU-CD datasets, surpassing existing comparative methods in terms of both precision metrics and computational efficiency. In addition, ablation experiments show that DHFCM and NAFRM are effective.
- 5 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 1-13
Top Cited PapersHA2F: Dual-Module Collaboration-Guided Hierarchical Adaptive Aggregation Framework for Remote Sensing Change Detection
Remote sensing change detection (RSCD) aims to identify the spatio-temporal changes of land cover, providing critical support for multi-disciplinary applications (e.g., environmental monitoring, disaster assessment, and climate change studies). Existing methods focus either on extracting features from localized patches, or pursue processing entire images holistically, which leads to the cross temporal feature matching deviation and exhibiting sensitivity to radiometric and geometric noise. Following the above issues, we propose a dual-module collaboration guided hierarchical adaptive aggregation framework, namely HA2F, which consists of dynamic hierarchical feature calibration module (DHFCM) and noise-adaptive feature refinement module (NAFRM). The former dynamically fuses adjacent-level features through perceptual feature selection, suppressing irrelevant discrepancies to address multi-temporal feature alignment deviations. The NAFRM utilizes the dual feature selection mechanism to highlight the change sensitive regions and generate spatial masks, suppressing the interference of irrelevant regions or shadows. Extensive experiments verify the effectiveness of the proposed HA2F, which achieves state-of-the-art performance on LEVIR-CD, WHU-CD, and SYSU-CD datasets, surpassing existing comparative methods in terms of both precision metrics and computational efficiency. In addition, ablation experiments show that DHFCM and NAFRM are effective.
- 19 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 1-15
DCCS-Det: Directional Context and Cross-Scale-Aware Detector for Infrared Small Target
Infrared small target detection (IRSTD) is critical for applications like remote sensing and surveillance, which aims to identify small, low-contrast targets against complex backgrounds. However, existing methods often struggle with inadequate joint modeling of local-global features (harming target-background discrimination) or feature redundancy and semantic dilution (degrading target representation quality). To tackle these issues, we propose DCCS-Det (Directional Context and Cross-Scale Aware Detector for Infrared Small Target), a novel detector that incorporates a Dual-stream Saliency Enhancement (DSE) block and a Latent-aware Semantic Extraction and Aggregation (LaSEA) module. The DSE block integrates localized perception with direction-aware context aggregation to help capture long-range spatial dependencies and local details. On this basis, the LaSEA module mitigates feature degradation via cross-scale feature extraction and random pooling sampling strategies, enhancing discriminative features and suppressing noise. Extensive experiments show that DCCS-Det achieves state-of-the-art detection accuracy with competitive efficiency across multiple datasets. Ablation studies further validate the contributions of DSE and LaSEA in improving target perception and feature representation under complex scenarios. Code is available at https://github.com/ML202010/DCCS-Det.
- 19 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 64 (01962892) , 1-15
DCCS-Det: Directional Context and Cross-Scale-Aware Detector for Infrared Small Target
Infrared small target detection (IRSTD) is critical for applications like remote sensing and surveillance, which aims to identify small, low-contrast targets against complex backgrounds. However, existing methods often struggle with inadequate joint modeling of local-global features (harming target-background discrimination) or feature redundancy and semantic dilution (degrading target representation quality). To tackle these issues, we propose DCCS-Det (Directional Context and Cross-Scale Aware Detector for Infrared Small Target), a novel detector that incorporates a Dual-stream Saliency Enhancement (DSE) block and a Latent-aware Semantic Extraction and Aggregation (LaSEA) module. The DSE block integrates localized perception with direction-aware context aggregation to help capture long-range spatial dependencies and local details. On this basis, the LaSEA module mitigates feature degradation via cross-scale feature extraction and random pooling sampling strategies, enhancing discriminative features and suppressing noise. Extensive experiments show that DCCS-Det achieves state-of-the-art detection accuracy with competitive efficiency across multiple datasets. Ablation studies further validate the contributions of DSE and LaSEA in improving target perception and feature representation under complex scenarios. Code is available at https://github.com/ML202010/DCCS-Det.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 23 (1545598X) , 1-5
MDAFNet: Multiscale Differential Edge and Adaptive Frequency Guided Network for Infrared Small Target Detection
Infrared small target detection (IRSTD) plays a crucial role in numerous military and civilian applications. However, the existing methods often face the gradual degradation of target edge pixels as the number of network layers increases, and traditional convolution struggles to differentiate between frequency components during feature extraction, leading to low-frequency backgrounds interfering with high-frequency targets and high-frequency noise triggering false detections. To address these limitations, we propose multiscale differential edge and adaptive frequency guided network for IRSTD (MDAFNet), which integrates the multiscale differential edge (MSDE) module and dual-domain adaptive feature enhancement (DAFE) module. The MSDE module, through a multiscale edge extraction and enhancement mechanism, effectively compensates for the cumulative loss of target edge information during downsampling. The DAFE module combines frequency-domain processing mechanisms with simulated frequency decomposition and fusion mechanisms in the spatial domain to effectively improve the network’s capability to adaptively enhance high-frequency targets and selectively suppress high-frequency noise. Experimental results on multiple datasets demonstrate the superior detection performance of MDAFNet.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 23 (1545598X) , 1-5
MDAFNet: Multiscale Differential Edge and Adaptive Frequency Guided Network for Infrared Small Target Detection
Infrared small target detection (IRSTD) plays a crucial role in numerous military and civilian applications. However, the existing methods often face the gradual degradation of target edge pixels as the number of network layers increases, and traditional convolution struggles to differentiate between frequency components during feature extraction, leading to low-frequency backgrounds interfering with high-frequency targets and high-frequency noise triggering false detections. To address these limitations, we propose multiscale differential edge and adaptive frequency guided network for IRSTD (MDAFNet), which integrates the multiscale differential edge (MSDE) module and dual-domain adaptive feature enhancement (DAFE) module. The MSDE module, through a multiscale edge extraction and enhancement mechanism, effectively compensates for the cumulative loss of target edge information during downsampling. The DAFE module combines frequency-domain processing mechanisms with simulated frequency decomposition and fusion mechanisms in the spatial domain to effectively improve the network’s capability to adaptively enhance high-frequency targets and selectively suppress high-frequency noise. Experimental results on multiple datasets demonstrate the superior detection performance of MDAFNet.
- 2 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. PP (01962892) , 1
Adaptive Frequency Separation Enhancement Network for Infrared Small Target Detection
Infrared small target detection aims to accurately extract small targets from complex backgrounds. It plays a crucial role in applications such as military reconnaissance and drone detection. Existing studies predominantly focus on spatial-domain feature extraction. However, they neglect to fully exploit frequency-domain information, which may contain critical patterns for comprehensive signal characterization. Consequently, this limitation results in suboptimal detection accuracy. To effectively capture the structural characteristics of small targets, frequency domain adaptive decomposition module (FDADM) is embedded into the spatial-domain encoder of the proposed network. The proposed FDADM effectively improves the accuracy of capturing target details and enhances structural feature consistency. Furthermore, considering that small targets may have blurred edges in complex backgrounds, gradient localization enhancement module is proposed. Gradient direction and gradient magnitude consistency functions are defined to precisely characterize target edges and contours, which improves detection accuracy. Experimental results demonstrate that the proposed method can effectively extract infrared small target under complex background conditions.
- 2 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. PP (01962892) , 1
Adaptive Frequency Separation Enhancement Network for Infrared Small Target Detection
Infrared small target detection aims to accurately extract small targets from complex backgrounds. It plays a crucial role in applications such as military reconnaissance and drone detection. Existing studies predominantly focus on spatial-domain feature extraction. However, they neglect to fully exploit frequency-domain information, which may contain critical patterns for comprehensive signal characterization. Consequently, this limitation results in suboptimal detection accuracy. To effectively capture the structural characteristics of small targets, frequency domain adaptive decomposition module (FDADM) is embedded into the spatial-domain encoder of the proposed network. The proposed FDADM effectively improves the accuracy of capturing target details and enhances structural feature consistency. Furthermore, considering that small targets may have blurred edges in complex backgrounds, gradient localization enhancement module is proposed. Gradient direction and gradient magnitude consistency functions are defined to precisely characterize target edges and contours, which improves detection accuracy. Experimental results demonstrate that the proposed method can effectively extract infrared small target under complex background conditions.
- 30 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-11
Global–Local Feature Collaborative Fusion and Edge Refinement Network for Remote Sensing Change Detection
Remote sensing image change detection (RSI-CD) has witnessed significant progress. The core focus lies in accurately identifying spatio-temporal changes in remote sensing images and precisely extracting edge details. However, current remote sensing image change detection primarily faces two challenges: the diversity and complexity of change regions, and the blurred edges of change regions. These issues lead to pseudo-change interference and incomplete detection of change targets. To alleviate these limitations, this study first develops a Local-Global Differential Fusion Module (LGDFM) that integrates local feature extraction with global self-attention mechanism to enhance change feature representation through effective bi-temporal image fusion. This module simultaneously captures both global and local information from source images, thereby overcoming the insufficient feature fusion in existing bi-temporal change detection methods. Subsequently, we propose an Edge-Aware Enhancement Module (EAEM) that employs morphological operations to extract edge features. The module progressively refines prediction results through multi-stage edge feature integration that ensures precise contour delineation. Experiments on two datasets demonstrate that the proposed method achieves competitive performance compared to existing approaches.
- 30 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-11
Global–Local Feature Collaborative Fusion and Edge Refinement Network for Remote Sensing Change Detection
Remote sensing image change detection (RSI-CD) has witnessed significant progress. The core focus lies in accurately identifying spatio-temporal changes in remote sensing images and precisely extracting edge details. However, current remote sensing image change detection primarily faces two challenges: the diversity and complexity of change regions, and the blurred edges of change regions. These issues lead to pseudo-change interference and incomplete detection of change targets. To alleviate these limitations, this study first develops a Local-Global Differential Fusion Module (LGDFM) that integrates local feature extraction with global self-attention mechanism to enhance change feature representation through effective bi-temporal image fusion. This module simultaneously captures both global and local information from source images, thereby overcoming the insufficient feature fusion in existing bi-temporal change detection methods. Subsequently, we propose an Edge-Aware Enhancement Module (EAEM) that employs morphological operations to extract edge features. The module progressively refines prediction results through multi-stage edge feature integration that ensures precise contour delineation. Experiments on two datasets demonstrate that the proposed method achieves competitive performance compared to existing approaches.
- Open Access
- 17 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 7480-7494
A Dual-Strategy Learning Framework for Hyperspectral Image Super-Resolution
Hyperspectral image super-resolution (HSI SR) has achieved remarkable success with deep neural networks. Currently, most methods in HSI SR assume a predetermined degradation model during training to synthesize low-resolution images (LR). These methods falter when confronted with HSI exhibiting degradation patterns and their limited flexibility restricts practical application. Additionally, these methods focus on the complex network designs for superior performance which entail high resource consumption and limit their broad application. To address these issues, in this paper, we propose a dual-strategy learning framework exploring meta-transfer learning for HSI blind SR. This framework can be applied to any SR network and facilitate performance enhancement. Firstly, we pretrain a three-channel SR model on natural image data to address the issue of insufficient HSI data. Furthermore, we innovatively propose a transfer scheme which directly applies our pretrained three-channel SR model to HSI, thereby significantly enhancing the spectral fidelity. To enhance the model's performance under specific degradation conditions, we incorporate meta-learning, enabling it to adapt to input images after a few iterations. Besides, we introduce attention-based knowledge distillation to equip our final network with the implicit representation capability of a meta network under a lightweight premise. Extensive experiments on three benchmark datasets demonstrate that the proposed approach outperforms existing methods in various degradations.
- 17 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 7480-7494
Open AccessA Dual-Strategy Learning Framework for Hyperspectral Image Super-Resolution
Hyperspectral image super-resolution (HSI SR) has achieved remarkable success with deep neural networks. Currently, most methods in HSI SR assume a predetermined degradation model during training to synthesize low-resolution images (LR). These methods falter when confronted with HSI exhibiting degradation patterns and their limited flexibility restricts practical application. Additionally, these methods focus on the complex network designs for superior performance which entail high resource consumption and limit their broad application. To address these issues, in this paper, we propose a dual-strategy learning framework exploring meta-transfer learning for HSI blind SR. This framework can be applied to any SR network and facilitate performance enhancement. Firstly, we pretrain a three-channel SR model on natural image data to address the issue of insufficient HSI data. Furthermore, we innovatively propose a transfer scheme which directly applies our pretrained three-channel SR model to HSI, thereby significantly enhancing the spectral fidelity. To enhance the model's performance under specific degradation conditions, we incorporate meta-learning, enabling it to adapt to input images after a few iterations. Besides, we introduce attention-based knowledge distillation to equip our final network with the implicit representation capability of a meta network under a lightweight premise. Extensive experiments on three benchmark datasets demonstrate that the proposed approach outperforms existing methods in various degradations.