A5060158251- Advanced SAR Imaging Techniques
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Microwave Imaging and Scattering Analysis
- Advanced Image Processing Techniques
- Advanced Neural Network Applications
- Digital Holography and Microscopy
- Sparse and Compressive Sensing Techniques
- Robotics and Sensor-Based Localization
- Advanced Vision and Imaging
- Advanced Optical Imaging Technologies
- Optical Coherence Tomography Applications
- Image Enhancement Techniques
- Image and Signal Denoising Methods
- Advanced Optical Sensing Technologies
- Video Surveillance and Tracking Methods
- Random lasers and scattering media
- Photoacoustic and Ultrasonic Imaging
- Optical measurement and interference techniques
- Geophysical Methods and Applications
- Image Processing Techniques and Applications
- Underwater Acoustics Research
- Infrared Target Detection Methodologies
- Ultrasonics and Acoustic Wave Propagation
- Nanoplatforms for cancer theranostics
- Medical Imaging Techniques and Applications
University of Electronic Science and Technology of China
2022-2025
Nanjing University of Aeronautics and Astronautics
2024-2025
National Institute for Materials Science
2024
University of Tsukuba
2024
University of Hong Kong
2018-2023
Guiyang Medical University
2020
SAR Ship Detection Dataset (SSDD) is the first open dataset that widely used to research state-of-the-art technology of ship detection from Synthetic Aperture Radar (SAR) imagery based on deep learning (DL). According our investigation, up 46.59% total 161 public reports confidently select SSDD study DL-based detection. Undoubtedly, this situation reveals popularity and great influence in remote sensing community. Nevertheless, coarse annotations ambiguous standards use its initial version...
Ship detection in synthetic aperture radar (SAR) images is a significant and challenging task. However, most existing deep learning-based SAR ship approaches are confined to single-polarization fail leverage dual-polarization characteristics, which increases the difficulty of further improving performance. One problem that requires solution how make full use characteristics excavate polarization features using network. To tackle problem, we propose group-wise feature enhancement-and-fusion...
Existing convolution neural network (CNN)-based video synthetic aperture radar (SAR) moving target shadow detectors are difficult to model long-range dependencies, while transformer-based ones often suffer from greater complexity. To handle these issues, this paper proposes MambaShadowDet, a novel lightweight deep learning (DL) detector based on state space (SSM), dedicated high-speed and high-accuracy detection in SAR images. By introducing SSM with the linear complexity into YOLOv8,...
Ship detection with rotated bounding boxes in synthetic aperture radar (SAR) images is now a hot spot. However, there are still some obstacles, such as multi-scale ships, misalignment between anchors and features, the opposite requirements for spatial sensitivity of regression tasks classification tasks. In order to solve these problems, we propose balanced feature-aligned network (RBFA-Net) where three targeted networks designed. They are, respectively, attention feature pyramid (BAFPN), an...
A capsule network, as an advanced technique in deep learning, is designed to overcome information loss the pooling operation and internal data representation of a convolutional neural network (CNN). It has shown promising results several applications, such digit recognition image segmentation. In this work, we investigate for first time use digital holographic reconstruction. The proposed residual encoder-decoder which call RedCap, uses novel windowed spatial dynamic routing algorithm block,...
Ship instance segmentation in synthetic aperture radar (SAR) images can provide more detailed location information and shape information, which is of great significance for port ship scheduling traffic management. However, there little research work on SAR segmentation, the general accuracy low because characteristics target task, such as multi-scale, aspect ratio, noise interference, are not considered. In order to solve these problems, we propose an idea scale (SIS) segmentation. Its...
Mask-based lensless imaging is an emerging modality, which replaces the lenses with optical elements and makes use of computation to reconstruct images from multiplexed measurements. Most existing reconstruction algorithms are implemented assuming that forward process a convolution operation, point spread function based on system model. In reality, there model mismatch, leading inferior image results. this paper, we investigate impact mismatch in mask-based for first time, illustrate...
Effective ship detection in synthetic aperture radar (SAR) imagery is crucial for maritime safety and surveillance. Despite the advancements deep learning SAR detection, significant challenges remain, particularly large scenes. These are twofold: of extremely small ships often hindered by inadequate feature extraction, presence inshore obscured pronounced land-based interference, both which lead to reduced accuracy. To address these issues, we present a novel framework that integrates...
Lensless imaging offers a lightweight, compact alternative to traditional lens-based systems, ideal for exploration in space-constrained environments. However, the absence of focusing lens and limited lighting such environments often results low-light conditions, where measurements suffer from complex noise interference due insufficient capture photons. This study presents robust reconstruction method high-quality scenarios, employing two complementary perspectives: model-driven data-driven....
Lensless imaging offers a lightweight, compact alternative to traditional lens-based systems, ideal for exploration in space-constrained environments. However, the absence of focusing lens and limited lighting such environments often result low-light conditions, where measurements suffer from complex noise interference due insufficient capture photons. This study presents robust reconstruction method high-quality scenarios, employing two complementary perspectives: model-driven data-driven....
We develop an image despeckling method that combines nonlocal self-similarity filters with machine learning, which makes use of convolutional neural network (CNN) denoisers. It consists three major steps: block matching, CNN despeckling, and group shrinkage. Through the we can take advantage similarity across patches as a regularizer to augment performance data-driven denoising using pre-trained network. The outputs from denoiser coordinates matching are further used form 3D groups similar...
The veiling effect caused by the scattering and absorption of suspending particles is a critical challenge underwater imaging. It possible to combine image formation model (IFM) with optical polarization characteristics effectively remove recover clear image. performance such methods, great extent, depends on settings global parameters in application scenarios. Meanwhile, learning-based methods can fit information degradation process nonlinearly restore images from scattering. Here, we...
In this study, we address the challenges associated with Video Synthetic Aperture Radar (Video SAR) shadow tracking, a technique used for continuous monitoring of ground moving targets. Due to such as changes in appearance, low contrast between and background, scene occlusion SAR, existing methods often encounter extensive matching errors data association process, resulting unsatisfactory tracking performance. To overcome these issues, propose novel method, GNN-JFL, which is based on joint...
Synthetic aperture radar (SAR) is an important active microwave imaging sensor [...]
Video synthetic aperture radar (Video SAR) has drawn much attention because it can continuously observe and track the moving target. Rather than tracking target directly, is better to its shadow no location shift, back-scattering characteristic stable. However, most current methods not only suffer from false alarms their discrimination capacities are good enough but also missed detection feature extraction limited under complicated environment. Therefore, we propose a shadow-enhanced...
A light field image captured by a plenoptic camera can be considered sampling of distribution within given space. However, with the limited pixel count sensor, acquisition high-resolution sample often comes at expense losing parallax information. In this work, we present learning-based generative framework to overcome such tradeoff directly simulating distribution. An important module our model is high-dimensional residual block, which fully exploits spatio-angular By learning distribution,...
Light field (LF) cameras often have significant limitations in spatial and angular resolutions due to their design. Many techniques that attempt reconstruct LF images at a higher resolution only consider either or resolution, but not both. We propose generative network using high-dimensional convolution improve both aspects. Our experimental results on synthetic real-world data demonstrate the proposed model outperforms existing state-of-the-art methods terms of peak signal-to-noise ratio...