A5100653198- Advanced Image Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Advanced Vision and Imaging
- Image Enhancement Techniques
- Advanced Image Fusion Techniques
- Image Processing Techniques and Applications
- Generative Adversarial Networks and Image Synthesis
- Data Management and Algorithms
- Video Surveillance and Tracking Methods
- Advanced Neural Network Applications
- Advanced Computational Techniques and Applications
- Medical Image Segmentation Techniques
- Human Pose and Action Recognition
- Rough Sets and Fuzzy Logic
- Stellar, planetary, and galactic studies
- Multimodal Machine Learning Applications
- Astro and Planetary Science
- Image and Signal Denoising Methods
- Adaptive optics and wavefront sensing
- Visual Attention and Saliency Detection
- Planetary Science and Exploration
- Image Processing and 3D Reconstruction
- Fire Detection and Safety Systems
- Service-Oriented Architecture and Web Services
Jinan University
2010-2025
Portland State University
2021-2024
Northwest University
2009-2023
Civil Aviation University of China
2020
China University of Petroleum, East China
2017
Beijing Normal University
2017
North-West University
2017
Brandeis University
2015-2016
Hainan University
2010
National University of Defense Technology
2010
This paper summarizes the 2nd NTIRE challenge on stereo image super-resolution (SR) with a focus new solutions and results. The task of is to super-resolve low-resolution pair high-resolution one magnification factor ×4. Compared single SR, major this lies in how exploit additional information another viewpoint maintain consistency has 3 tracks, including track distortion (e.g., PSNR) bicubic degradation, perceptual quality LPIPS) as well real degradations. In total, 175, 93, 103...
We present Hunyuan3D 2.0, an advanced large-scale 3D synthesis system for generating high-resolution textured assets. This includes two foundation components: a shape generation model -- Hunyuan3D-DiT, and texture Hunyuan3D-Paint. The generative model, built on scalable flow-based diffusion transformer, aims to create geometry that properly aligns with given condition image, laying solid downstream applications. benefiting from strong geometric priors, produces vibrant maps either generated...
In this paper, a transmission-guided lightweight neural network called TGL-Net is proposed for efficient image dehazing. Unlike most current dehazing methods that produce simulated transmission maps from depth data and haze-free images, in the work, guided are computed automatically using filter-refined dark-channel-prior (F-DCP) method real-world hazy images as regularizer, which facilitates training not only on synthetic data, but also natural images. A double-error loss function combines...
Unsupervised image-to-image translation (UI2I) tasks aim to find a mapping between the source and target domains from unpaired training data. Previous methods can not effectively capture differences domain on different scales often leads poor quality of generated images, noise, distortion, other conditions that do match human vision perception, has high time complexity. To address this problem, we propose multi-scale structure progressive growth generator method solve UI2I task. Our refines...
Image haze removal techniques have a promising application prospects. However, color distortion by excessive dehazing is common problem especially for images containing light surface objects. To solve this problem, an improved image method based on dark channel prior with fidelity proposed in paper. First, filters (Sobel operator and mean filter) are used to refine the initial transmission map estimated from prior. Second, piecewise constrained function keep fidelity. Experimental results...
Novel view synthesis of dynamic scenes has been an intriguing yet challenging problem. Despite recent advancements, simultaneously achieving high-resolution photorealistic results, real-time rendering, and compact storage remains a formidable task. To address these challenges, we propose Spacetime Gaussian Feature Splatting as novel scene representation, composed three pivotal components. First, formulate expressive Gaussians by enhancing 3D with temporal opacity parametric motion/rotation....
Chinese Opera LianPu (COLP) is an essential intangible cultural heritage that embodies profound historical, cultural, and artistic significance. The digitization of COLP not only holds significant practical value but also carries the responsibility preserving inheriting traditional culture. This paper aims to establish a Image Dataset (COLP-ID) provide solid foundation for further research in this art form. Additionally, Multi-scale Attentional Feature Fusion model(MSAFF) proposed. It...
Web mining is an important sub-branch of data mining. The technology normally adopts integration method to generate warehouse, on which dig the relation rules, cluster characters and get useful module prediction knowledge evaluation. For data's semi-structure heterogeneous (mixed media) character, different pure based database. In this paper we introduce a multi-agent cooperation module, play as crawler, exaction machine, generalization machine analysis machine. Through communication between...
This study assesses the outcomes of NTIRE 2023 Challenge on Non-Homogeneous Dehazing, wherein novel techniques were proposed and evaluated new image dataset called HD-NH-HAZE. The HD-NH-HAZE contains 50 high resolution pairs real-life outdoor images featuring nonhomogeneous hazy corresponding haze-free same scene. haze was simulated using a professional setup that replicated real-world conditions scenarios. competition had 246 participants 17 teams submitted solutions for final testing...
Disentanglement research is a critical and important issue in the field of image editing. In order to perform disentangled editing on images generated by generative models, this paper presents an unsupervised, model-agnostic, two-stage trained framework. This work addresses problem discovering interpretable, directions edited attributes latent space models. effort’s primary objective was address limitations discovered previous research, mainly (a) are interpretable but significantly...
In a car license plate recognition system, effective and robust image expansion methods will improve its performance bring lower error rate. Two MAP-based super-resolution reconstruction approaches for single with prior model described as Huber Markov random field are discussed applied to such system in this paper. A new spatial smoothness measurement based on flexible convolution kernel is proposed. Parameters these discussed. Improved definition of images increased rate also shown through...
Data analysts often engage in data exploration tasks to discover interesting patterns, without knowing exactly what they are looking for. Such can be very labor-intensive because require the user review many results of ad-hoc queries and adjust predicates subsequent balance tradeoff between collecting all information reducing size returned data. In this demonstration we introduce AIDE , a system that automates these tasks. steers towards areas based on her relevance feedback database...
Query optimizers have long been considered as among the most complex components of a database engine, while assessment an optimizer's quality remains challenging task. Indeed, existing performance benchmarks for engines (like TPC benchmarks) produce query runtime system rather than its optimizer. To address this challenge, paper introduces OptMark, toolkit evaluating OptMark is designed to offer number desirable properties. First, it decouples optimizer from underlying execution engine....
In literatures, there were several forms of definitions for temporal association rules which have different formulas calculating support and confidence. this paper, after reviewing the we first reform definition rule, then give a general form rule was equipped with new formula its Based on definition, propose fast algorithm mining rules. Experiments synthetic dataset real exhibit good performance proposed algorithm.
Edge detection is often performed on disc-like object in Cassini astronomy images to accurately obtain the object’s center position. The existing edge extraction methods usually produce lots of false pixels because noise and interior details objects. In paper, an algorithm based Extreme Learning Machine (ELM) proposed for images. ELM model, a 28-D feature vector pixel image constructed as input, which consists first second derivatives some Haar-like features, binary classifier obtained...