A5027177558- Remote-Sensing Image Classification
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Remote Sensing and Land Use
- Advanced SAR Imaging Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Advanced Image Fusion Techniques
- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Domain Adaptation and Few-Shot Learning
- Robotics and Sensor-Based Localization
- Sparse and Compressive Sensing Techniques
- Robotic Path Planning Algorithms
- Advanced Vision and Imaging
- Image Processing Techniques and Applications
- Monoclonal and Polyclonal Antibodies Research
- Remote Sensing and LiDAR Applications
- Soil Moisture and Remote Sensing
- Robotic Locomotion and Control
- Infrared Target Detection Methodologies
- Medical Image Segmentation Techniques
- Speech and Audio Processing
- Venous Thromboembolism Diagnosis and Management
- Crystal Structures and Properties
- Water Systems and Optimization
Chinese Academy of Sciences
2025
Shanghai Institute of Materia Medica
2025
Henan University
2025
Nanjing University of Aeronautics and Astronautics
2014-2024
Beijing Tian Tan Hospital
2024
Shandong University
2021-2022
Guidewire (United States)
2022
Cytoskeleton (United States)
2022
Shandong University of Science and Technology
2021
Northwestern Polytechnical University
2017
To solve the problems such as obvious speckle noise and serious spectral distortion when existing fusion methods are applied to of optical SAR images, this paper proposes a method for images based on Dense-UGAN Gram–Schmidt transformation. Firstly, dense connection with U-shaped network (Dense-UGAN) used in GAN generator deepen structure obtain deeper source image information. Secondly, according particularity imaging mechanism, SGLCM loss preserving texture features PSNR reducing introduced...
Synthetic Aperture Radar (SAR) can provide rich feature information under all-weather and day-night conditions because it is not affected by climatic conditions. However, multiplicative speckle noise exists in SAR images, which makes difficult to accurately identify some fuzzy targets such as roads rivers, during semantic segmentation. This paper proposes an improved Deeplabv3+ network that be effectively applied the segmentation task of images. Firstly, this added attention mechanism and,...
Synthetic aperture radar (SAR) provides rich information about the Earth’s surface under all-weather and day-and-night conditions, is applied in many relevant fields. SAR imagery semantic segmentation, which can be a final product for end users fundamental procedure to support other applications, one of most difficult challenges. This paper proposes an encoding-decoding network based on Deeplabv3+ semantically segment imagery. A new potential energy loss function Gibbs distribution proposed...
Deep-learning-based methods have obtained satisfying results in polarimetric synthetic aperture radar (PolSAR) image classification. However, these require large numbers of labeled samples, which are usually time-consuming and high-priced for PolSAR images. To address this issue, a semisupervised method based on 3-D convolutional neural network (3-D-CNN) using pseudo labels (PL-3-D-CNN) is proposed. First, the coherency matrix data converted into 6-D real-valued vector by unitary...
ABSTRACTABSTRACTSynthetic aperture radar (SAR) has all-weather and all-day observation capabilities a certain ability to penetrate the surface; therefore, it unique advantages in many aspects, which other remote sensing methods cannot achieve. However, owing imaging principles, interpretation of SAR images is complicated. Therefore, converting into optical one method image interpretation. this paper proposes an improved conditional generative adversarial network (cGAN) that includes...
The purpose of remote sensing image fusion is to synthesize the characteristics multi-source images and generate a composite with new spatial, spectral temporal features. Gram-Schmidt (GS) transform can better improve spatial features maintain original large extent. However, for Synthetic aperture radar (SAR) optical image, there are still obvious distortion detail blurring. In this paper, GS method improved non-subsampled contourlet (NSCT) which used obtain high-resolution that contains...
Synthetic Aperture Radar (SAR) imagery is significant in remote sensing, but the limited spatial resolution results restricted detail and clarity. Current super-resolution methods confront challenges such as complex network structure, insufficient sensing capability, difficulty extracting features with local global dependencies. To address these challenges, DMSC-GAN, a SAR image technique based on c-GAN framework, introduced this study. The design objective of DMSC-GAN to enhance flexibility...
Detecting crack type and size is crucial for road maintenance management. Mobile crowd sensing a new way to collect the information of cracks on roads. We propose system named CrackDetector detect estimate their types with smart phone in this paper. The (i.e., horizontal crack, vertical net crack) determined by coordinate transmission method based three directions, including direction photo (estimated through an image processing method), 3D shooting rotation smartphone while photographing)...
Unmanned aerial vehicles (UAV) have had significant progress in the last decade, applying to many fields for its convenience explore areas that men cannot reach and of image processing. Still, as basis further application, semantic segmentation is one most difficult challenges. In this paper, we propose a method urban UAV images segmentation, utilizing geographical information, digital surface models (DSM). We introduce an end-to-end, dual stream fully convolutional networks (FCN) based...
Evaluation of similarity between simulated synthetic aperture radar (SAR) images and real SAR is crucial to improve simulating algorithms it still lacks quantitative evaluation images. Thus, this paper proposes four methods (i.e., PCA similarity, SIFT matching rate, change detection rate difference-value target field view) evaluate the Experimental results indicate that can be applied river, plain terrain, mountainous terrain man-made target, while view only applicable for target.
Unmanned aerial vehicles (UAV) have had significant progress in the last decade, which is applied to many relevant fields because of image processing and convenience explore areas that men cannot reach. Still, as basis further applications such object tracking terrain classification, semantic segmentation one most difficult challenges field computer vision. In this paper, we propose a method for urban UAV images segmentation, utilizes geographical information region interest form digital...
In view of the redundancy features in feature level fusion PolSAR and optical images, Support Vector Machine Recursive Feature Elimination (SVM-RFE) algorithm is adopted. However, due to low efficiency for a large number datasets, an improved screening proposed. Firstly, extract various characteristics image image, then through LASSO construct penalty function delete redundant features, combining with SVM-RFE method, get best sub sets; Secondly, by random forest conditional field, it...
Superresolution mapping (SRM) is a technology to handle mixed pixels in remote sensing image. In this letter, novel hybrid interpolation-based SRM by parallel paths (HISRM-PP) proposed. Firstly, the two different high resolution fractional images for each class are respectively derived paths. Then kinds of integrated produce higher appropriate weighting parameter. Finally, utilized obtain result allocation. Due HISRM-PP, more spatial-spectral information original image improve accuracy...
In terms of land cover classification, optical images have been proven to good classification performance. Synthetic Aperture Radar (SAR) has the characteristics working all-time and all-weather. It more significant advantages over for recognition some scenes, such as water bodies. One current challenges is how fuse benefits both obtain powerful capabilities. This study proposes a model based on random forest with conditional fields (CRF) feature-level fusion using features extracted from...
In order to solve the problem of motion planning with dynamics, RRT algorithm was proposed. But environment becoming complex, tree exploration slows down significantly. this paper, a novel two-stage is The first stage discrete search based on scent pervasion principle, and second sampling-based under direction information. results show good performance convergence speed proposed algorithm.
A novel two-stage gridless sparsity-based method for sparse linear array direction-of-arrival (DOA) estimation is presented. First, the covariance matrix representation based on atomic norm minimisation proposed corresponding structured Toeplitz construction and source number detection. Then, conventional MUSIC algorithm can be employed DOA estimation. Compared with subspace-based algorithms, carried out without knowing directly detect more signals than sensors. Numerical simulations...
Structure from Motion (SfM) has been proved an efficient algorithm of 3-D point cloud reconstruction derived optical images. This paper extends it to infrared images taken by thermal cameras. To solve the absence distinctive features and presence reflections with low contrast, this proposed a new TAC-RANSAC model eliminate mismatches using feature detection suitable for The experiment shows that method reduced number obtain ideal result reconstruction.
This paper proposes a new method of restoration and segmentation SAR image. The radar cross-section (RCS) for intensity images is estimated based on Gibbs Markov random fields simulated annealing. Further, it puts forward to segment image into target shadow with the theory connectivity in digital morphology. In this paper, field models annealing used together processing are discussed first time; simple effective presented using connected cluster model pixels value relevance neighborhood...
With the rapid development of modern world, it is imperative to achieve effective and efficient monitoring for territories interest, especially broad ocean area. For surveillance ship targets at sea, a common powerful approach take advantage satellite synthetic aperture radar (SAR) systems. Currently, using SAR images classification challenging issue due complex sea situations imaging variances ships. Fortunately, emergence advanced sensors has shed much light on automatic target recognition...
<title>Abstract</title> <bold>Objective</bold>To explore the incidence and risk factors for deep vein thrombosis (DVT) pulmonary embolism (PE) following surgical intervention meningioma. <bold>Methods</bold> In this retrospective, observational study, we enrolled 9067 patients with histologically confirmed meningiomas who underwent resection at our institution between January 2019 June 2024. Demographic data (including gender, age, geographic region) information on postoperative...