A5048621299- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Image Processing Techniques and Applications
- Advanced Image and Video Retrieval Techniques
- Visual Attention and Saliency Detection
- Image Enhancement Techniques
- Image and Signal Denoising Methods
- Advanced Image Fusion Techniques
- Face recognition and analysis
- Video Coding and Compression Technologies
- Polymer composites and self-healing
- Multimodal Machine Learning Applications
- AI in cancer detection
- Image and Video Quality Assessment
- Radiomics and Machine Learning in Medical Imaging
- Advanced Fluorescence Microscopy Techniques
- Generative Adversarial Networks and Image Synthesis
- Domain Adaptation and Few-Shot Learning
- Video Surveillance and Tracking Methods
- Cell Image Analysis Techniques
- Esophageal Cancer Research and Treatment
- Computer Graphics and Visualization Techniques
- Lung Cancer Diagnosis and Treatment
- Retinal Imaging and Analysis
- Corrosion Behavior and Inhibition
Fudan University
2016-2025
Texas A&M University
2024
Yanshan University
2024
Southwest University
2011-2023
Guangdong Academy of Medical Sciences
2023
Coatings Research Institute
2020
Nanjing Tech University
2009-2016
Chongqing University
2013-2014
Universiti Teknologi Petronas
2014
Guangzhou Medical University
2010-2013
No-reference image quality assessment (NR-IQA) is a non-trivial task, because it hard to find pristine counterpart for an in real applications, such as selection, high recommendation, etc. In recent years, deep learning-based NR-IQA methods emerged and achieved better performance than previous methods. this paper, we present novel neural networks-based multi-task learning approach NR-IQA. Our proposed network designed by manner that consists of two tasks, namely, natural scene statistics...
Identifying small size images or objects is a notoriously challenging problem, as discriminative representations are difficult to learn from the limited information contained in them with poor-quality appearance and unclear object structure. Existing research works usually increase resolution of low-resolution image pixel space order provide better visual quality for human viewing. However, improved performance such methods even trivial case very (we will show it this paper explicitly). In...
Single-image super-resolution (SISR) is a classic problem in the image processing community, which aims at generating high-resolution from low-resolution one. In recent years, deep learning based SISR methods emerged and achieved performance leap than previous methods. However, because evaluation metrics of peak signal-to-noise ratio (PSNR), usually choose L2-norm as loss function. This leads to significant improvement final PSNR value but little perceptual quality. this paper, order achieve...
Pictures taken under the low-light condition often suffer from low contrast and loss of image details, thus an approach that can effectively improve images is demanded. Traditional Retinex-based methods assume reflectance components keep unchanged, which neglect color distortion lost details. In this paper, we propose end-to-end learning-based framework first decomposes then learns to fuse decomposed results obtain high quality enhanced result. Our be divided into a RDNet (Retinex...
Melamine formaldehyde (MF) resins have been synthesized at different reaction temperature and pH values. Different molar ratios of melamine were used to synthesize the corresponding resins. The prepared resin samples characterized by using molecular weight determination viscometry thermogravimetric analysis (TGA). maximum percentage solid content (69.7%) was obtained 8.5 75°C temperature. MF increased with an increase monomer concentration. highest residual 14.125 wt.% sample 10.
Supervised deep learning techniques have achieved great success in various fields due to getting rid of the limitation handcrafted representations. However, most previous image retargeting algorithms still employ fixed design principles such as using gradient map or features compute saliency map, which inevitably restricts its generality. Deep may help address this issue, but challenging problem is that we need build a large-scale dataset for training models. building requires huge human...
Under stereo settings, the problems of disparity estimation, magnification and stereo-view synthesis have gathered wide attention. However, limited image quality brings non-negligible difficulties in developing related applications becomes main bottleneck images. To best our knowledge, restoration is rarely studied. Towards this end, paper analyses how to effectively explore information, proposes a unified framework. The proposed framework explicitly learn inherent pixel correspondence...
Background Non-magnifying endoscopy with narrow-band imaging (NM-NBI) has been frequently used in routine screening of esophagus squamous cell carcinoma (ESCC). The performance NBI for early ESCC is, however, significantly affected by operator experience. Artificial intelligence may be a unique approach to compensate the lack Aim To construct computer-aided detection (CAD) system application NM-NBI identify and compare it our previously reported CAD endoscopic white-light (WLI). Methods A...
Abstract Optical neural networks (ONNs) are particularly advantageous owing to their inherent parallelism and low energy consumption. However, one of the obstacles implementation ONNs is lack optical nonlinearity. In this study, nonlinear activators for prepared by combining Ti 3 C 2 T x MXene with microfibers principles verified. Activation functions obtained from experimental measurements used simulate multiclassification super‐resolution reconstruction tasks performance comparable that...
Abstract In this work, an active protective epoxy coating with weathering resistant, corrosion-warning, and self-healing properties was developed by incorporating tannic acid (TA) loaded mesoporous silica (MSN-TA) nanocontainers. The introduction of MSN-TA nanocontainers could alleviate the degradation via scavenging radicals generated during UV irradiation. Compared blank coating, containing 5 wt.% exhibited much less in surface morphology, wettability glossiness, maintained a good barrier...
For video super-resolution, current state-of-the-art approaches either process multiple low-resolution (LR) frames to produce each output high-resolution (HR) frame separately in a sliding window fashion or recurrently exploit the previously estimated HR super-resolve following frame. The main weaknesses of these are: 1) generating may obtain high-quality estimates while resulting unsatisfactory flickering artifacts, and 2) combining generated can temporally consistent results case short...
Metal–organic frameworks (MOFs) for enzyme encapsulation induced biomimetic mineralization are commonly microporous and hydrophobic, which result in a rather high mass transfer resistance of the reactants restrain catalytic activity.
Image-based single-modality compression learning approaches have demonstrated exceptionally powerful encoding and decoding capabilities in the past few years , but suffer from blur severe semantics loss at extremely low bitrates. To address this issue, we propose a multimodal machine method for text-guided image compression, which semantic information of text is used as prior to guide better performance. We fully study role description different components codec, demonstrate its...
With the sharp increasing of mobile devices, conducting search on devices becomes pervasive, and one most popular applications is visual search. To achieve low bit-rate search, existing works focus addressing local descriptor coding BoW histogram compression . In this paper, we extend concept image retargeting propose a new resizing approach that devoted to preserving robust features in query while it. Based extended concept, introduce novel mobile-visual-search scheme conducts proposed...