A5108303648- Advanced Steganography and Watermarking Techniques
- Digital Media Forensic Detection
- Image Processing Techniques and Applications
- Video Surveillance and Tracking Methods
- Autonomous Vehicle Technology and Safety
- Advanced Neural Network Applications
- Seismic Imaging and Inversion Techniques
- Hydrocarbon exploration and reservoir analysis
- Advanced Text Analysis Techniques
- Gait Recognition and Analysis
- NMR spectroscopy and applications
- Human Pose and Action Recognition
Hangzhou Dianzi University
2024-2025
Beijing Technology and Business University
2022
ABSTRACT Deep learning methods for image forgery detection often struggle with compression attack robustness. This paper proposes a novel multi‐class framework combining spatial‐frequency fusion Swin‐Transformer, outperforming existing in scenarios. Our approach integrates frequency domain perception module quantization tables, spatial through multi‐strategy convolutions, and dual‐attention mechanism channel attention feature fusion. Experimental results demonstrate superior performance an F...
The majority of deep learning methods for detecting image forgery fail to accurately detect and localize the tampering operations. Furthermore, they only support a single type. Our method introduces three key innovations: (1) A spatial perception module that combines rich model (SRM) with constrained convolution, enabling focused detection traces while suppressing interference from content; (2) hierarchical feature architecture integrates Swin Transformer UperNet effective multi-scale...
Abstract Although graph convolutional networks have achieved good performances in skeleton‐graph‐based action recognition, there are still some problems which include the incomplete utilization of skeleton features and lacking logical adjacency information between nodes matrix. In this article, a human recognition algorithm is proposed based on multiple from to solve these problems. More specifically, an improved matrix constructed make full use features. These local differential features,...
Summary Continuous wavelet transform (CWT) as a seismic time-frequency analysis technique with multi-resolution characteristics has been widely used in interpretation, such hydrocarbon detection. With the increased degree of oil and gas exploration, exploration targets are gradually shifting to complex reservoirs lithologic structural-lithologic reservoirs. In this case, CWT method become increasingly unable meet accuracy resolution requirements detection due Heisenberg uncertainty...