- Synthetic Aperture Radar (SAR) Applications and Techniques
- Advanced SAR Imaging Techniques
- Underwater Acoustics Research
- Ocean Waves and Remote Sensing
- Data Mining Algorithms and Applications
- Remote Sensing and Land Use
- Rough Sets and Fuzzy Logic
- Remote-Sensing Image Classification
- Advanced Neural Network Applications
- Automated Road and Building Extraction
- Software System Performance and Reliability
- Data Quality and Management
- Arctic and Antarctic ice dynamics
- Soil, Finite Element Methods
- Geophysical Methods and Applications
- Image Processing Techniques and Applications
- Service-Oriented Architecture and Web Services
- Metaheuristic Optimization Algorithms Research
- Engineering Structural Analysis Methods
- Data Stream Mining Techniques
- Human Pose and Action Recognition
- Advanced Image Fusion Techniques
- Remote Sensing and LiDAR Applications
- Identification and Quantification in Food
- Web Data Mining and Analysis
Beijing University of Posts and Telecommunications
2021-2024
Northwestern Polytechnical University
2022-2024
Beijing Academy of Artificial Intelligence
2023
China Telecom (China)
2023
China Telecom
2023
Hubei University of Technology
2021
Wuhan Technology and Business University
2019
Tsinghua University
2015-2017
Tongji University
2016
Sichuan Institute of Building Research
2015
This paper proposes a coastline detection method for polarimetric synthetic aperture radar (SAR) images based on region-based and edge-based active contour models. It can be used to detect accurately fast. In this method, the models are effectively combined by an important property of likelihood ratio edge detector in SAR images, which is proved theory. Using low-resolution image obtained multilook processing, we accurate continued coarse coastlines level set method. The along region then...
The difficulty of coastline detection in polarimetric SAR images is that the coast zone includes both sea noised by sidelobe echo strong scattering targets and intertidal zones with varying water content. Methods based solely on statistical distribution quad-polarization coherent matrix or single-channel intensity would result incorrect segmentation. In this paper, a method three-region Markov random field (MRF) segmentation embedded components obtained Freeman decomposition proposed....
In object detection, offset-guided and point-guided regression dominate anchor-based anchor-free method separately. Recently, approach is introduced to method. However, we observe points predicted by this way are misaligned with matched region of proposals score localization, causing a notable gap in performance. paper, propose CPM R-CNN which contains three efficient modules optimize anchor- based According sufficient evaluations on the COCO dataset, demonstrated improve localization...
Offshore oil platforms are difficult to detect due the complex sea state, sparseness of target distribution, and similarity targets with ships. In this paper, we propose an platform detection method in polarimetric synthetic aperture radar (PolSAR) images using level set segmentation a limited initial region convolutional neural network (CNN). Firstly, reduce interference clutter, offshore strong scattering were initially detected by generalized optimization contrast enhancement (GOPCE)...
Abstract As a novel swarm intelligence optimization algorithm, cuckoo search (CS) has been successfully applied to solve diverse problems in the real world. Despite its efficiency and wide use, CS some disadvantages, such as premature convergence, easy fall into local optimum poor balance between exploitation exploration. In order improve performance of new extension with multi-swarms Q-Learning namely MP-QL-CS is proposed. The step size strategy algorithm that an individual fitness value...
One key problem for the classification of multi-frequency polarimetric SAR images is to extract target features simultaneously in aspects frequency, polarization and spatial texture. This paper proposes a new method data based on tensor representation multi-linear subspace learning (MLS). Firstly, each cell represented by third-order domains, with order corresponding one domain. Then, two main MLS methods, i.e., principal component analysis (MPCA) extension linear discriminant (MLDA), are...
To address the problem of threshold segmentation polarimetric images in coastal zones when intensity is not bimodal under complex environments, a thresholding method based on three-component decomposition and likelihood ratio proposed this article. First, double-bounce volume scattering powers are extracted by decomposition. Then, three-step diagram carried out two powers. The base point determined sampling typical window regions with minimum or maximum average power variance first. interval...
It is difficult to detect bridges in synthetic aperture radar (SAR) images due the inherent speckle noise of SAR images, interference generated by strong coastal scatterers, and diversity bridge terrain morphologies. In this paper, we present a two-step detection method for polarimetric imagery, which probability graph model Markov tree used build water network, are detected traversing network determine all adjacent branch pairs. step construction, candidate branches first extracted using...
It is difficult to detect ports in polarimetric SAR images due the complicated components, morphology, and coastal environment. This paper proposes an unsupervised port detection method by extracting water of based on three-component decomposition multi-scale thresholding segmentation. Firstly, characteristics are analyzed using modified decomposition. Secondly, volume scattering power ratio double-bounce (PRDV) used extract water. Water land first separated a global segmentation power,...
As sea-crossing bridges are important hubs connecting separated land areas, their detection in SAR images is of great significance. However, under complex scenarios, the sea surface conditions, distribution coastal terrain morphologies, and scattering components different structures bridge area very diverse, which makes accurate robust difficult, including sea–land segmentation feature extraction on depends. In this paper, we propose a polarimetric image method for based windowed level set...
The detection of harbors presents difficulties related to their diverse sizes, varying morphology and scattering, complex backgrounds. To avoid the extraction unstable geometric features, in this paper, we propose an unsupervised harbor method for polarimetric SAR images using context features reflection symmetry. First, image is segmented into three region types, i.e., water low-scattering regions, strong-scattering urban other based on a multi-region Markov random field (MRF) segmentation...
Abstract Human matting refers to extracting human parts from natural images with high quality, including detail information such as hair, glasses, hats, etc. This technology plays an essential role in image synthesis and visual effects the film industry. When green screen is not available, existing methods need help of additional inputs (such trimap, background image, etc.), or model computational cost complex network structure, which brings great difficulties application practice. To...
This article proposes a symmetric sparse representation (SSR) method to extract pure endmembers from hyperspectral imagery (HSI). The SSR combines the features of linear unmixing model and subspace clustering endmembers, it assumes that desired all HSI pixel points can be sparsely represented by each other. It formulates endmember extraction problem into famous program archetypal analysis, accordingly, extracting transformed as finding archetypes in minimal convex hull containing points....
AIOps (Artificial Intelligence for IT Operations) has emerged as a powerful solution to tackle the challenges involved in operating and maintaining complex microservice systems. Inspired by AIOps, this study proposes novel approach that leverages knowledge graph construct an intelligent operation maintenance (O&M) system. We first called OpsKG, then realize series of functions based on OpsKG O&M system, including alarm query, root-cause analysis location, prediction. With practical...
Hearing aids have become an indispensable part of the lives some hearing-impaired people. Traditional hearing will be adjusted according to personal curve and allowing patients avoid noise-induced harm. However, there is no sound classification or intelligent noise reduction, which cannot meet higher demand for aids. This paper designed a aid based on two-level neural network, Urbansound8K data set was used train network. It can simulate human auditory attention mechanism intelligently...
For the high time overhead problems of Apriori algorithm while solving for long length frequent patterns, using MapReduce distributed programming ideas, paper breaks original idea Aproiri which discovers item sets through gradually increasing element numbers in sets. It proposes a new non-iteration parallel about pattern discovery, can get arbitrary at random. The experimental results show that proposed has better performance than such algorithms are under ideas traditional algorithm.
As many large organizations have multiple data sources and the scale of dataset becomes larger larger, it is inevitable to carry out mining in distributed environment. In this paper, we address problem global frequent closed itemsets A novel algorithm proposed obtain with exact frequency shown that can determine all itemsets. new structure developed maintain Then an efficient implementation provided based on structure. Experimental results show effective.