Su Qiu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Su Qiu has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
- Weiqi Jin
Beijing Institute of Technology
15 shared publications Weiqi Jin
MOE Key Laboratory of Optoelectronic Imaging Technology and Systems Beijing Institute of Technology 100081 Beijing China
5 shared publicationsXia Wang
MOE Key Laboratory of Optoelectronic Imaging Technology and Systems Beijing Institute of Technology 100081 Beijing China
4 shared publicationsJiaan Xue
MOE Key Laboratory of Optoelectronic Imaging Technology and Systems Beijing Institute of Technology 100081 Beijing China
3 shared publications- Jie Yang
MOE Key Laboratory of Optoelectronic Imaging Technology and System, Beijing Institute of Technology, Beijing 100081, China
3 shared publications - Taoran Lu
MOE Key Laboratory of Optoelectronic Imaging Technology and System, Beijing Institute of Technology, Beijing 100081, China
3 shared publications
Latest Publications
- Open Access
- 10 May 2025
- Published by MDPI AG in Remote Sensing
- Vol. 17 (10) , 1683
Multimodal Prompt-Guided Bidirectional Fusion for Referring Remote Sensing Image Segmentation
Multimodal feature alignment is a key challenge in referring remote sensing image segmentation (RRSIS). The complex spatial relationships and multi-scale targets in remote sensing images call for efficient cross-modal mapping and fine-grained feature alignment. Existing approaches typically rely on cross-attention for multimodal fusion, which increases model complexity. To address this, we introduce the concept of prompt learning in RRSIS and propose a parameter-efficient multimodal prompt-guided bidirectional fusion (MPBF) architecture. MPBF combines both early and late fusion strategies. In the early fusion stage, it conducts the deep fusion of linguistic and visual features through cross-modal prompt coupling. In the late fusion stage, to handle the multi-scale nature of remote sensing targets, a scale refinement module is proposed to capture diverse scale representations, and a vision–language alignment module is employed for pixel-level multimodal semantic associations. Comparative experiments and ablation studies on a public dataset demonstrate that MPBF significantly outperformed existing state-of-the-art methods with relatively small computational overhead, highlighting its effectiveness and efficiency for RRSIS. Further application experiments on a custom dataset confirm the method’s practicality and robustness in real-world scenarios.
- 10 May 2025
- Published by MDPI AG in Remote Sensing
- Vol. 17 (10) , 1683
Open AccessMultimodal Prompt-Guided Bidirectional Fusion for Referring Remote Sensing Image Segmentation
Multimodal feature alignment is a key challenge in referring remote sensing image segmentation (RRSIS). The complex spatial relationships and multi-scale targets in remote sensing images call for efficient cross-modal mapping and fine-grained feature alignment. Existing approaches typically rely on cross-attention for multimodal fusion, which increases model complexity. To address this, we introduce the concept of prompt learning in RRSIS and propose a parameter-efficient multimodal prompt-guided bidirectional fusion (MPBF) architecture. MPBF combines both early and late fusion strategies. In the early fusion stage, it conducts the deep fusion of linguistic and visual features through cross-modal prompt coupling. In the late fusion stage, to handle the multi-scale nature of remote sensing targets, a scale refinement module is proposed to capture diverse scale representations, and a vision–language alignment module is employed for pixel-level multimodal semantic associations. Comparative experiments and ablation studies on a public dataset demonstrate that MPBF significantly outperformed existing state-of-the-art methods with relatively small computational overhead, highlighting its effectiveness and efficiency for RRSIS. Further application experiments on a custom dataset confirm the method’s practicality and robustness in real-world scenarios.
- Open Access
Dual-Encoder UNet-Based Narrowband Uncooled Infrared Imaging Denoising Network
Uncooled infrared imaging systems have significant potential in industrial hazardous gas leak detection. However, the use of narrowband filters to match gas spectral absorption peaks leads to a low level of incident energy captured by uncooled infrared cameras. This results in a mixture of fixed pattern noise and Gaussian noise, while existing denoising methods for uncooled infrared images struggle to effectively address this mixed noise, severely hindering the extraction and identification of actual gas leak plumes. This paper presents a UNet-structured dual-encoder denoising network specifically designed for narrowband uncooled infrared images. Based on the distinct characteristics of Gaussian random noise and row–column stripe noise, we developed a basic scale residual attention (BSRA) encoder and an enlarged scale residual attention (ESRA) encoder. These two encoder branches perform noise perception and encoding across different receptive fields, allowing for the fusion of noise features from both scales. The combined features are then input into the decoder for reconstruction, resulting in high-quality infrared images. Experimental results demonstrate that our method effectively denoises composite noise, achieving the best results according to both objective metrics and subjective evaluations. This research method significantly enhances the signal-to-noise ratio of narrowband uncooled infrared images, demonstrating substantial application potential in fields such as industrial hazardous gas detection, remote sensing imaging, and medical imaging.
Open AccessDual-Encoder UNet-Based Narrowband Uncooled Infrared Imaging Denoising Network
Uncooled infrared imaging systems have significant potential in industrial hazardous gas leak detection. However, the use of narrowband filters to match gas spectral absorption peaks leads to a low level of incident energy captured by uncooled infrared cameras. This results in a mixture of fixed pattern noise and Gaussian noise, while existing denoising methods for uncooled infrared images struggle to effectively address this mixed noise, severely hindering the extraction and identification of actual gas leak plumes. This paper presents a UNet-structured dual-encoder denoising network specifically designed for narrowband uncooled infrared images. Based on the distinct characteristics of Gaussian random noise and row–column stripe noise, we developed a basic scale residual attention (BSRA) encoder and an enlarged scale residual attention (ESRA) encoder. These two encoder branches perform noise perception and encoding across different receptive fields, allowing for the fusion of noise features from both scales. The combined features are then input into the decoder for reconstruction, resulting in high-quality infrared images. Experimental results demonstrate that our method effectively denoises composite noise, achieving the best results according to both objective metrics and subjective evaluations. This research method significantly enhances the signal-to-noise ratio of narrowband uncooled infrared images, demonstrating substantial application potential in fields such as industrial hazardous gas detection, remote sensing imaging, and medical imaging.
- 19 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 62 (01962892) , 1-15
Multiscale Spatial-Frequency Domain Dynamic Pansharpening of Remote Sensing Images Integrated With Wavelet Transform
Pansharpening produces high-resolution multi-spectral (HRMS) images by enhancing the high-frequency details in multispectral (MS) images, with deep learning (DL) methods emerging as a new paradigm. However, most existing DL methods primarily focus on pansharpening in the spatial domain using convolutional layers with local kernels, which is not conducive to accurate reconstruction of global high frequencies, leading to spatial deformation and spectral distortion. To address this issue, we propose a bidirectional progressive spatial-frequency dynamic fusion network that constructs a multi-scale structure using wavelet transform. This network utilizes the wavelet transform’s ability to analyze frequency information in local neighborhoods to inject high frequencies from panchromatic (PAN) images into MS images. Additionally, it employs a wavelet-based adaptive fusion module to analyze spatial-spectral correlations and adaptively learn the injection gains for high-frequency details, thereby enhancing spectral fidelity. To further adapt to various scenarios, a dual-domain dynamic filtering block is proposed. It improves model generalization through data-specific dynamic filtering in both frequency and spatial domains, capturing global and local dependencies effectively. Quantitative metrics and visualization results on reduced and full-resolution images demonstrate that our method outperforms other advanced methods with low computational cost, producing high-fidelity HRMS images.
- 19 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 62 (01962892) , 1-15
Multiscale Spatial-Frequency Domain Dynamic Pansharpening of Remote Sensing Images Integrated With Wavelet Transform
Pansharpening produces high-resolution multi-spectral (HRMS) images by enhancing the high-frequency details in multispectral (MS) images, with deep learning (DL) methods emerging as a new paradigm. However, most existing DL methods primarily focus on pansharpening in the spatial domain using convolutional layers with local kernels, which is not conducive to accurate reconstruction of global high frequencies, leading to spatial deformation and spectral distortion. To address this issue, we propose a bidirectional progressive spatial-frequency dynamic fusion network that constructs a multi-scale structure using wavelet transform. This network utilizes the wavelet transform’s ability to analyze frequency information in local neighborhoods to inject high frequencies from panchromatic (PAN) images into MS images. Additionally, it employs a wavelet-based adaptive fusion module to analyze spatial-spectral correlations and adaptively learn the injection gains for high-frequency details, thereby enhancing spectral fidelity. To further adapt to various scenarios, a dual-domain dynamic filtering block is proposed. It improves model generalization through data-specific dynamic filtering in both frequency and spatial domains, capturing global and local dependencies effectively. Quantitative metrics and visualization results on reduced and full-resolution images demonstrate that our method outperforms other advanced methods with low computational cost, producing high-fidelity HRMS images.
- Open Access
Underwater Single-Photon 3D Reconstruction Algorithm Based on K-Nearest Neighbor
The high sensitivity and picosecond time resolution of single-photon avalanche diodes (SPADs) can improve the operational range and imaging accuracy of underwater detection systems. When an underwater SPAD imaging system is used to detect targets, backward-scattering caused by particles in water often results in the poor quality of the reconstructed underwater image. Although methods such as simple pixel accumulation have been proven to be effective for time–photon histogram reconstruction, they perform unsatisfactorily in a highly scattering environment. Therefore, new reconstruction methods are necessary for underwater SPAD detection to obtain high-resolution images. In this paper, we propose an algorithm that reconstructs high-resolution depth profiles of underwater targets from a time–photon histogram by employing the K-nearest neighbor (KNN) to classify multiple targets and the background. The results contribute to the performance of pixel accumulation and depth estimation algorithms such as pixel cross-correlation and ManiPoP. We use public experimental data sets and underwater simulation data to verify the effectiveness of the proposed algorithm. The results of our algorithm show that the root mean square errors (RMSEs) of land targets and simulated underwater targets are reduced by 57.12% and 23.45%, respectively, achieving high-resolution single-photon depth profile reconstruction.
Open AccessUnderwater Single-Photon 3D Reconstruction Algorithm Based on K-Nearest Neighbor
The high sensitivity and picosecond time resolution of single-photon avalanche diodes (SPADs) can improve the operational range and imaging accuracy of underwater detection systems. When an underwater SPAD imaging system is used to detect targets, backward-scattering caused by particles in water often results in the poor quality of the reconstructed underwater image. Although methods such as simple pixel accumulation have been proven to be effective for time–photon histogram reconstruction, they perform unsatisfactorily in a highly scattering environment. Therefore, new reconstruction methods are necessary for underwater SPAD detection to obtain high-resolution images. In this paper, we propose an algorithm that reconstructs high-resolution depth profiles of underwater targets from a time–photon histogram by employing the K-nearest neighbor (KNN) to classify multiple targets and the background. The results contribute to the performance of pixel accumulation and depth estimation algorithms such as pixel cross-correlation and ManiPoP. We use public experimental data sets and underwater simulation data to verify the effectiveness of the proposed algorithm. The results of our algorithm show that the root mean square errors (RMSEs) of land targets and simulated underwater targets are reduced by 57.12% and 23.45%, respectively, achieving high-resolution single-photon depth profile reconstruction.
- Open Access
A Simulation Method for Underwater SPAD Depth Imaging Datasets
In recent years, underwater imaging and vision technologies have received widespread attention, and the removal of the backward-scattering interference caused by impurities in the water has become a long-term research focus for scholars. With the advent of new single-photon imaging devices, single-photon avalanche diode (SPAD) devices, with high sensitivity and a high depth resolution, have become cutting-edge research tools in the field of underwater imaging. However, the high production costs and small array areas of SPAD devices make it very difficult to conduct underwater SPAD imaging experiments. To address this issue, we propose a fast and effective underwater SPAD data simulation method and develop a denoising network for the removal of backward-scattering interference in underwater SPAD images based on deep learning and simulated data. The experimental results show that the distribution difference between the simulated and real underwater SPAD data is very small. Moreover, the algorithm based on deep learning and simulated data for the removal of backward-scattering interference in underwater SPAD images demonstrates effectiveness in terms of both metrics and human observation. The model yields improvements in metrics such as the PSNR, SSIM, and entropy of 5.59 dB, 9.03%, and 0.84, respectively, demonstrating its superior performance.
Open AccessA Simulation Method for Underwater SPAD Depth Imaging Datasets
In recent years, underwater imaging and vision technologies have received widespread attention, and the removal of the backward-scattering interference caused by impurities in the water has become a long-term research focus for scholars. With the advent of new single-photon imaging devices, single-photon avalanche diode (SPAD) devices, with high sensitivity and a high depth resolution, have become cutting-edge research tools in the field of underwater imaging. However, the high production costs and small array areas of SPAD devices make it very difficult to conduct underwater SPAD imaging experiments. To address this issue, we propose a fast and effective underwater SPAD data simulation method and develop a denoising network for the removal of backward-scattering interference in underwater SPAD images based on deep learning and simulated data. The experimental results show that the distribution difference between the simulated and real underwater SPAD data is very small. Moreover, the algorithm based on deep learning and simulated data for the removal of backward-scattering interference in underwater SPAD images demonstrates effectiveness in terms of both metrics and human observation. The model yields improvements in metrics such as the PSNR, SSIM, and entropy of 5.59 dB, 9.03%, and 0.84, respectively, demonstrating its superior performance.
- Open Access
Automatic Suppression Method for Water Surface Glints Using a Division of Focal Plane Visible Polarimeter
To address the problem of water surface detection imaging equipment being susceptible to water surface glints, this study demonstrates a method called De-Glints for suppressing glints and obtaining clear underwater images using a division of focal plane (DoFP) polarimeter. Based on the principle of polarization imaging, the best polarization angle and the image corresponding to the minimal average gray level of each pixel are calculated. To evaluate the improvement in image quality, the index E was designed. The results of indoor and outdoor experiments show that the error of the angle calculation of this method is within 10%, and the minimum error is only 3%. The E index is positively improved and can be relatively improved by 8.00 under the interference of strong outdoor glints, and the method proposed in this paper shows a good adaptive ability to the dynamic scene.
Open AccessAutomatic Suppression Method for Water Surface Glints Using a Division of Focal Plane Visible Polarimeter
To address the problem of water surface detection imaging equipment being susceptible to water surface glints, this study demonstrates a method called De-Glints for suppressing glints and obtaining clear underwater images using a division of focal plane (DoFP) polarimeter. Based on the principle of polarization imaging, the best polarization angle and the image corresponding to the minimal average gray level of each pixel are calculated. To evaluate the improvement in image quality, the index E was designed. The results of indoor and outdoor experiments show that the error of the angle calculation of this method is within 10%, and the minimum error is only 3%. The E index is positively improved and can be relatively improved by 8.00 under the interference of strong outdoor glints, and the method proposed in this paper shows a good adaptive ability to the dynamic scene.
- Open Access
- 8 August 2023
- Published by MDPI AG in Remote Sensing
- Vol. 15 (16) , 3933
Color-Coated Steel Sheet Roof Building Extraction from External Environment of High-Speed Rail Based on High-Resolution Remote Sensing Images
The identification of color-coated steel sheet (CCSS) roof buildings in the external environment is of great significance for the operational security of high-speed rail systems. While high-resolution remote sensing images offer an efficient approach to identify CCSS roof buildings, achieving accurate extraction is challenging due to the complex background in remote sensing images and the extensive scale range of CCSS roof buildings. This research introduces the deformation-aware feature enhancement and alignment network (DFEANet) to address these challenges. DFEANet adaptively adjusts the receptive field to effectively separate the foreground and background facilitated by the deformation-aware feature enhancement module (DFEM). Additionally, feature alignment and gated fusion module (FAGM) is proposed to refine boundaries and preserve structural details, which can ameliorate the misalignment between adjacent features and suppress redundant information during the fusion process. Experimental results on remote sensing images along the Beijing–Zhangjiakou high-speed railway demonstrate the effectiveness of DFEANet. Ablation studies further underscore the enhancement in extraction accuracy due to the proposed modules. Overall, the DFEANet was verified as capable of assisting in the external environment security of high-speed rails.
- 8 August 2023
- Published by MDPI AG in Remote Sensing
- Vol. 15 (16) , 3933
Open AccessColor-Coated Steel Sheet Roof Building Extraction from External Environment of High-Speed Rail Based on High-Resolution Remote Sensing Images
The identification of color-coated steel sheet (CCSS) roof buildings in the external environment is of great significance for the operational security of high-speed rail systems. While high-resolution remote sensing images offer an efficient approach to identify CCSS roof buildings, achieving accurate extraction is challenging due to the complex background in remote sensing images and the extensive scale range of CCSS roof buildings. This research introduces the deformation-aware feature enhancement and alignment network (DFEANet) to address these challenges. DFEANet adaptively adjusts the receptive field to effectively separate the foreground and background facilitated by the deformation-aware feature enhancement module (DFEM). Additionally, feature alignment and gated fusion module (FAGM) is proposed to refine boundaries and preserve structural details, which can ameliorate the misalignment between adjacent features and suppress redundant information during the fusion process. Experimental results on remote sensing images along the Beijing–Zhangjiakou high-speed railway demonstrate the effectiveness of DFEANet. Ablation studies further underscore the enhancement in extraction accuracy due to the proposed modules. Overall, the DFEANet was verified as capable of assisting in the external environment security of high-speed rails.
- Open Access
- 13 July 2023
- Published by Optica Publishing Group in Optica Open
Self-calibration algorithm for installation angle deviation of bionic polarization compound eyes
A self-calibration algorithm based on unsupervised optimization for polarizer installation angle deviation is proposed and used in a multi-aperture bionic polarization compound eye system. To simplify calibration operation, under the condition that the calibration-polarized light information is unknown, this algorithm fully exploits redundancy and random polarization information in the scene, and uses a non-convex multi-objective discrete parameter sorting optimization method to achieve angle self-calibration. Compared with ordinary calibration procedures, the algorithm requires less stringent conditions, achieves online calibration and is more accurate. It also can be applied to camera polarization arrays, division-of-focal-plane polarization cameras, and other polarization devices.
- 13 July 2023
- Published by Optica Publishing Group in Optica Open
Open AccessSelf-calibration algorithm for installation angle deviation of bionic polarization compound eyes
A self-calibration algorithm based on unsupervised optimization for polarizer installation angle deviation is proposed and used in a multi-aperture bionic polarization compound eye system. To simplify calibration operation, under the condition that the calibration-polarized light information is unknown, this algorithm fully exploits redundancy and random polarization information in the scene, and uses a non-convex multi-objective discrete parameter sorting optimization method to achieve angle self-calibration. Compared with ordinary calibration procedures, the algorithm requires less stringent conditions, achieves online calibration and is more accurate. It also can be applied to camera polarization arrays, division-of-focal-plane polarization cameras, and other polarization devices.
- Open Access
- 13 July 2023
- Published by Optica Publishing Group in Optica Open
Self-calibration algorithm for installation angle deviation of bionic polarization compound eyes
A self-calibration algorithm based on unsupervised optimization for polarizer installation angle deviation is proposed and used in a multi-aperture bionic polarization compound eye system. To simplify calibration operation, under the condition that the calibration-polarized light information is unknown, this algorithm fully exploits redundancy and random polarization information in the scene, and uses a non-convex multi-objective discrete parameter sorting optimization method to achieve angle self-calibration. Compared with ordinary calibration procedures, the algorithm requires less stringent conditions, achieves online calibration and is more accurate. It also can be applied to camera polarization arrays, division-of-focal-plane polarization cameras, and other polarization devices.
- 13 July 2023
- Published by Optica Publishing Group in Optica Open
Open AccessSelf-calibration algorithm for installation angle deviation of bionic polarization compound eyes
A self-calibration algorithm based on unsupervised optimization for polarizer installation angle deviation is proposed and used in a multi-aperture bionic polarization compound eye system. To simplify calibration operation, under the condition that the calibration-polarized light information is unknown, this algorithm fully exploits redundancy and random polarization information in the scene, and uses a non-convex multi-objective discrete parameter sorting optimization method to achieve angle self-calibration. Compared with ordinary calibration procedures, the algorithm requires less stringent conditions, achieves online calibration and is more accurate. It also can be applied to camera polarization arrays, division-of-focal-plane polarization cameras, and other polarization devices.
- Open Access
- 8 July 2023
- Published by Optica Publishing Group in Optica Open
Self-calibration algorithm for installation angle deviation of bionic polarization compound eyes
A self-calibration algorithm based on unsupervised optimization for polarizer installation angle deviation is proposed and used in a multi-aperture bionic polarization compound eye system. To simplify calibration operation, under the condition that the calibration-polarized light information is unknown, this algorithm fully exploits redundancy and random polarization information in the scene, and uses a non-convex multi-objective discrete parameter sorting optimization method to achieve angle self-calibration. Compared with ordinary calibration procedures, the algorithm requires less stringent conditions, achieves online calibration and is more accurate. It also can be applied to camera polarization arrays, division-of-focal-plane polarization cameras, and other polarization devices.
- 8 July 2023
- Published by Optica Publishing Group in Optica Open
Open AccessSelf-calibration algorithm for installation angle deviation of bionic polarization compound eyes
A self-calibration algorithm based on unsupervised optimization for polarizer installation angle deviation is proposed and used in a multi-aperture bionic polarization compound eye system. To simplify calibration operation, under the condition that the calibration-polarized light information is unknown, this algorithm fully exploits redundancy and random polarization information in the scene, and uses a non-convex multi-objective discrete parameter sorting optimization method to achieve angle self-calibration. Compared with ordinary calibration procedures, the algorithm requires less stringent conditions, achieves online calibration and is more accurate. It also can be applied to camera polarization arrays, division-of-focal-plane polarization cameras, and other polarization devices.