A5101464630- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Sparse and Compressive Sensing Techniques
- Remote-Sensing Image Classification
- Antenna Design and Analysis
- Advanced Image Fusion Techniques
- Advanced Neural Network Applications
- Image Processing Techniques and Applications
- Multimodal Machine Learning Applications
- Animal Disease Management and Epidemiology
- Infrared Target Detection Methodologies
- Advanced Antenna and Metasurface Technologies
- Influenza Virus Research Studies
- Viral Infections and Vectors
- Advanced Data Compression Techniques
- Metamaterials and Metasurfaces Applications
- Microwave Imaging and Scattering Analysis
- Human Motion and Animation
- Video Analysis and Summarization
Peking University
2015-2024
Guangxi University
2024
Chinese Academy of Sciences
2022
Shanghai Institute of Technical Physics
2022
University of Chinese Academy of Sciences
2022
Harbin Institute of Technology
2015
Block transform coded images usually suffer from annoying artifacts at low bit-rates, because of the independent quantization DCT coefficients. Image prior models play an important role in compressed image reconstruction. Natural patches a small neighborhood high-dimensional space exhibit underlying sub-manifold structure. To model distribution signal, we extract structure as knowledge. We utilize graph Laplacian regularization to characterize patch level. And similar are exploited samples...
Recent deep models trained on large-scale RGB datasets lead to considerable achievements in visual detection tasks. However, the training examples are often limited for an infrared task, which may deteriorate performance of detectors. In this paper, we propose a transfer approach, Source Model Guidance (SMG), where leverage high-capacity model as guidance supervise process network. SMG, foreground soft label generated from is introduced source knowledge provide cross-domain transfer....
Deep neural networks are widely used for image restoration, however the loss criteration is usually set as ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> . penalizes larger errors, which unstable outliers. To avoid disadvantages, xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> utilized a more robust and well behaved loss. This paper proposes novel function restoration networks, measures geodesic distance in Riemannian manifold...
Sparsity has shown promising results in various image restoration applications. Recent advances have suggested that structured or group sparsity often leads to more powerful compression artifact reduction studies. In this paper, we introduce nonlocal multi-dimension an adaptive space-transform domain, which performs multi-scale wavelet transform on DCT coefficients of similar patches. The new efficiently reduces redundancies between inner block and inter simultaneously, thus it can...
Equine influenza (EI) is a severe infectious disease that causes huge economic losses to the horse industry. Spatial epidemiology technology can explore spatiotemporal distribution characteristics and occurrence risks of diseases, it has played an important role in prevention control major diseases humans animals. For first time, this study conducted systematic analysis EI using SaTScan software investigated environmental variables suitable areas for Maxent model. A total 517 occurrences...
This paper proposes a method to estimate coefficients for blocking artifact reduction at low bit rate. Across-resolution coherence that and high resolution image are similar is introduced preserve signal continuity. Non-local similarity used provide samples estimation by searching blocks of reference block. We have two sources estimation. One source exploiting non-local decoded image, interpolating the resolution. obtain based on across different resolutions. The other quantization...
We propose a flexible and fast estimation method to calculate the far-field patterns of digital-coding metasurfaces (DCMs) by performing chirp Z-transform (CZT), called DCM-CZT method. Because expression form convolution, CZT can be accelerated Fourier transform. Compared with traditional discrete transform (DFT) method, accurately estimate arbitrary element periods. More importantly, partial for some specific orientations, instead global like DFT does. show that efficiently improve...
The theory of compressive sensing (CS) has attracted considerable research interests from signal and image processing communities. And in practice, because the considerations data storage transmission, scalar quantization is necessary to be implemented on CS measurements. In this paper, we propose an adaptive bandwise sparsity regularization handle recovery problem quantized sensing. constraints every patch by using distribution model transform domain. addition, bring cost function quantify...
Total-variation (TV) regularization is widely adopted in image restoration problems to exploit the local smoothness of image. However, traditional TV only models sparsity gradient at original scale. This paper introduces a multi-scale method which different scales, and constrains magnitude scales jointly. As extract frequency image, our proposed provides constraints for components. And each scale, we adaptively estimate distribution particular pixel from group nonlocally searched similar...