A5100445482- Image Enhancement Techniques
- Advanced Image Processing Techniques
- Visual Attention and Saliency Detection
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
- Domain Adaptation and Few-Shot Learning
- Image Processing Techniques and Applications
- Image and Signal Denoising Methods
- Multimodal Machine Learning Applications
- Sparse and Compressive Sensing Techniques
- Digital Media Forensic Detection
- Advanced Steganography and Watermarking Techniques
- Advanced Image Fusion Techniques
- Face and Expression Recognition
- Video Surveillance and Tracking Methods
- Image Retrieval and Classification Techniques
- Face Recognition and Perception
- Human Pose and Action Recognition
- Industrial Vision Systems and Defect Detection
- Advanced Data Compression Techniques
- Wireless Communication Networks Research
- Rough Sets and Fuzzy Logic
- Blind Source Separation Techniques
- Video Coding and Compression Technologies
Tianjin University
2016-2025
China Coal Technology and Engineering Group Corp (China)
2024-2025
China Coal Research Institute (China)
2025
Tiandi Science & Technology (China)
2025
Beijing Academy of Artificial Intelligence
2023-2024
Shanghai Artificial Intelligence Laboratory
2023-2024
Nanchang University
2022
The paper presents a novel method, Zero-Reference Deep Curve Estimation (Zero-DCE), which formulates light enhancement as task of image-specific curve estimation with deep network. Our method trains lightweight network, DCE-Net, to estimate pixel-wise and high-order curves for dynamic range adjustment given image. is specially designed, considering pixel value range, monotonicity, differentiability. Zero-DCE appealing in its relaxed assumption on reference images, i.e., it does not require...
Images captured under water are usually degraded due to the effects of absorption and scattering. Degraded underwater images show some limitations when they used for display analysis. For example, with low contrast color cast decrease accuracy rate object detection marine biology recognition. To overcome those limitations, a systematic image enhancement method, which includes an dehazing algorithm algorithm, is proposed. Built on minimum information loss principle, effective proposed restore...
Underwater vision suffers from severe effects due to selective attenuation and scattering when light propagates through water. Such degradation not only affects the quality of underwater images but limits ability tasks. Different existing methods which either ignore wavelength dependency or assume a specific spectral profile, we tackle color distortion problem image new view. In this letter, propose weakly supervised transfer method correct distortion, relaxes need paired for training allows...
Rapid development of affordable and portable consumer depth cameras facilitates the use information in many computer vision tasks such as intelligent vehicles 3D reconstruction. However, map captured by low-cost sensors (e.g., Kinect) usually suffers from low spatial resolution, which limits its potential applications. In this paper, we propose a novel deep network for super-resolution (SR), called DepthSR-Net. The proposed DepthSR-Net automatically infers high-resolution (HR) low-resolution...
During recent years, we have witnessed a rapid development of wireless network technologies which revolutionized the way people take and share multimedia content. However, images captured in outdoor scenes usually suffer from limited visibility due to suspended atmospheric particles, directly affects quality photos. Despite progress image dehazing methods, visual dehazed results still needs further improvement. In this paper, propose deep convolutional neural (CNN) for single called PDR-Net,...
Images captured under outdoor scenes usually suffer from low contrast and limited visibility due to suspended atmospheric particles, which directly affects the quality of photographs. Despite numerous image dehazing methods have been proposed, effective hazy restoration remains a challenging problem. Existing learning-based predict medium transmission by convolutional neural networks (CNNs), but ignore key global light. Different previous methods, we propose flexible cascaded CNN for single...
This work reviews the results of NTIRE 2021 Challenge on Non-Homogeneous Dehazing. The proposed techniques and their have been evaluated a novel dataset that extends NH-Haze datset. It consists additional 35 pairs real haze free nonhomogeneous hazy images recorded outdoor. has introduced in outdoor scenes by using professional setup imitates conditions scenes. 327 participants registered challenge 23 teams competed final testing phase. solutions gauge state-of-the-art image dehazing.
Underwater object detection is a crucial and challenging problem in marine engineering aquatic robotics. The difficulty partly because of the degradation underwater images caused by light selective absorption scattering. Intuitively, enhancing can benefit high-level applications like detection. However, it still unclear whether all detectors need image enhancement as preprocessing. We therefore pose questions <italic xmlns:mml="http://www.w3.org/1998/Math/MathML"...
As an important and challenging problem in computer vision, zero-shot learning (ZSL) aims at automatically recognizing the instances from unseen object classes without training data. To address this problem, ZSL is usually carried out following two aspects: 1) capturing domain distribution connections between seen data 2) modeling semantic interactions image feature space label embedding space. Motivated by these observations, we propose a bidirectional mapping-based relationship scheme that...
Poor visibility due to the effects of light absorption and scattering is challenging for processing underwater images. We propose an approach based on dehazing color correction algorithms image enhancement. First, a simple algorithm applied remove haze in image. Second, compensation, histogram equalization, saturation, intensity stretching are used improve contrast, brightness, color, Furthermore, bilateral filtering utilized address problem noise caused by physical properties medium...
Zero-shot learning (ZSL) is typically achieved by resorting to a class semantic embedding space transfer the knowledge from seen classes unseen ones. Capturing common characteristics between visual modality and (e.g., attributes or word vector) key success of ZSL. In this paper, we propose novel encoder-decoder approach, namely latent encoding (LSE), connect relations different modalities. Instead requiring projection function information across modalities like most previous work, LSE...
Restoring underwater image from a single is known to be an ill-posed problem. Some assumptions made in previous methods are not suitable many situations. In this paper, effective method proposed restore images. Using the quad-tree subdivision and graph-based segmentation, global background light can robustly estimated. The medium transmission map estimated based on minimum information loss principle optical properties of imaging. Qualitative experiments show that our results characterized by...
Several metrics have been proposed in literature to assess the quality of 2D images, but devoted assessment stereoscopic images are very scarce. Therefore, this paper, an objective method is predict level images. This assesses stereo from perspective image and sense. Experiments demonstrate that paper presented gets similar results with general subjective method. And simple, rapid, convenient practical.
Zero-shot learning (ZSL) endows the computer vision system with inferential capability to recognize new categories that have never seen before. Two fundamental challenges in it are visual-semantic embedding and domain adaptation cross-modality unseen class prediction steps, respectively. This paper presents two corresponding methods named Adaptive STructural Embedding (ASTE) Self-PAced Selective Strategy (SPASS) for both challenges. Specifically, ASTE formulates interactions a latent...
Abstract Unmanned aerial vehicles (UAVs) are widely used in wireless communication networks due to their rapid deployment and high mobility. However, practical scenarios, the existence of obstacles eavesdroppers will seriously interfere with quality UAV network produce a security risk. Thus, this paper combines reconfigurable intelligent surface (RIS) technology UAVs build secure network. Normally, rotary‐wing (labeled as UAV‐S) acting base station sends information signals legitimate user...