A5038677687- Remote Sensing and Land Use
- Automated Road and Building Extraction
- Advanced Optical Imaging Technologies
- 3D Modeling in Geospatial Applications
- Spam and Phishing Detection
- Network Security and Intrusion Detection
- Image Retrieval and Classification Techniques
- Image Processing and 3D Reconstruction
- Advanced Glycation End Products research
- Video Surveillance and Tracking Methods
- Advanced Memory and Neural Computing
- Phytochemicals and Antioxidant Activities
- Neural Networks and Reservoir Computing
- Tea Polyphenols and Effects
- Geological Modeling and Analysis
- Imbalanced Data Classification Techniques
Tianjin University of Science and Technology
2024
Shanghai Polytechnic University
2023
Guilin University of Technology
2020
The detection of fraud accounts at scenarios registration and login is a critical task for Internet enterprises, which can help to avoid economic losses very early stage. In industry, most companies tend deploy supervised models such as rule-based models. However, these methods are significantly restricted in scalability since they highly rely on domain knowledge manual annotations. Therefore, we designed novel account graph analysis approach uncovering fraudulent patterns.
Convolutional neural networks (CNNs) are representative models of artificial (ANNs), that form the backbone modern computer vision. However, considerable power consumption and limited computing speed electrical platforms restrict further development CNNs. Optical considered next-generation physical implementations ANNs to break bottleneck. This study proposes a spectral convolutional network (SCNN) with function matter meta-imaging, namely identifying composition mapping its distribution in...
Abstract. Although contemporary geospatial science has made great progress, spatial data fusion of vector and raster is still a problem in the geoinformation environment. In order to solve problem, this paper proposes method which merges data. Firstly, row column numbers data, X, Y values are represented by Morton code C++ environment, respectively. Secondly, we establish table Oracle database store Third, uses minimum selection bounding box extract top building model. Finally, divide into...
Abstract. There are some problems in the visualization of 3D model buildings at present, such as redundant texture data and large memory occupation loading data, which pose challenges to smooth building visualization. In this paper, we propose a method using fractal quadtrees simplify compress management. Firstly, uses self-similarity feature that same has dimension screen all textures, when multiple textures is within threshold range. Radon transform will be performed on these calculate...
Abstract. Shadows are ubiquitous in high-resolution images, especially urban regions where there more serious shadow occlusions. In order to improve the detection effect of shadows, this paper analyzes characteristics and properties shadows orthophotos, proposes an orthophoto method under artificial shadow. Firstly, modeling tool is used calculate (i.e. shadow) caused by building obstructing sun's rays. Secondly, relaxation matching algorithm extended position shape polygon as characteristic...