A5100772205- Advanced Vision and Imaging
- Advanced Image and Video Retrieval Techniques
- Generative Adversarial Networks and Image Synthesis
- Image Retrieval and Classification Techniques
- Image Enhancement Techniques
- Face and Expression Recognition
- Advanced Image Processing Techniques
- 3D Shape Modeling and Analysis
- Image Processing Techniques and Applications
- Robotics and Sensor-Based Localization
- Video Surveillance and Tracking Methods
- Image and Signal Denoising Methods
- Computer Graphics and Visualization Techniques
- Sparse and Compressive Sensing Techniques
- Medical Image Segmentation Techniques
- Multimodal Machine Learning Applications
- Video Analysis and Summarization
- Face recognition and analysis
- Advanced Measurement and Detection Methods
- Image and Video Stabilization
- Optical measurement and interference techniques
- Optical Systems and Laser Technology
- Additive Manufacturing and 3D Printing Technologies
- 3D Surveying and Cultural Heritage
- Industrial Vision Systems and Defect Detection
Nanchang Hangkong University
2014-2024
Capital Normal University
2024
China State Shipbuilding (China)
2024
Wuhan University of Science and Technology
2012-2023
Guangxi Normal University
2023
Guilin University of Aerospace Technology
2023
Wuhu Hit Robot Technology Research Institute
2022
China University of Geosciences
2021
University of Science and Technology of China
2021
Guilin University of Electronic Technology
2020
Web caching or web proxy has been considered as the prime vehicle of coping with ever-increasing demand for information retrieval over Internet, WWW being a typical example. Existing work on primarily focused content based caching; relatively less attention given to development proper placement strategies potential proxies in Internet. In this paper, we argue that is critical performance and further investigates optimal policy target server The objective optimize measure subject system...
Image-to-image translation is an important and challenging problem in computer vision image processing. Diffusion models (DM) have shown great potentials for high-quality synthesis, gained competitive performance on the task of image-to-image translation. However, most existing diffusion treat as conditional generation processes, suffer heavily from gap between distinct domains. In this paper, a novel method based Brownian Bridge Model (BBDM) proposed, which stochastic process, learns two...
Image registration is an important technique in many computer vision applications such as image fusion, retrieval, object tracking, face recognition, change detection and so on. Local feature descriptors, i.e., how to detect features describe them, play a fundamental role process, which directly influence the accuracy robustness of registration. This paper mainly focuses on variety local descriptors including some theoretical research, mathematical models, methods or algorithms along with...
Image colorization aims to produce a natural looking color image from given gray-scale image, which remains challenging problem. In this paper, we propose novel example-based method exploiting new locality consistent sparse representation. Given single reference our automatically colorizes the target by pursuit. For efficiency and robustness, operates at superpixel level. We extract low-level intensity features, mid-level texture high-level semantic features for each superpixel, are then...
Given a reference color image and destination grayscale image, this paper presents novel automatic colorization algorithm that transfers information from the to image. Since images may contain content at different or even varying scales (due changes of distance between objects camera), existing texture matching-based methods can often perform poorly. We propose cross-scale matching method improve robustness quality results. Suitable are considered locally, which then fused using global...
In stereo vision-based three-dimensional measurements, calibration, and matching are the most challenging tasks for accurate reconstruction. The traditional binocular vision algorithm has low precision, we propose a system using telecentric lens. this paper, calibration system, which collects only two pictures of plate to complete calibration. is special application 3D if measurement points selected as on grid, appropriate extension can be used high-precision reconstruction, will explained...
In this paper, we revisit the manifold assumption which has been widely adopted in learning-based image super-resolution. The states that point-pairs from high-resolution share local geometry with corresponding low-resolution manifold. However, does not hold always, since one-to-multiple mapping LR to HR makes neighbor reconstruction ambiguous and results blurring artifacts. To minimize ambiguous, utilize Locality Preserving Constraints (LPC) avoid confusions through emphasizing consistency...
ABSTRACT Optical pooled screening (OPS) is a highly scalable method for linking image-based phenotypes with cellular perturbations. However, it has thus far been restricted to relatively low-plex phenotypic readouts in cancer cell lines culture, due limitations associated situ sequencing (ISS) of perturbation barcodes. Here, we developed PerturbView, an OPS technology that leverages vitro transcription (IVT) amplify barcodes prior ISS, enabling screens multiplexed across diverse systems,...
The performance of human listeners degrades rather slowly compared to machines in noisy environments. This has been attributed the ability performing auditory scene analysis which separates speech prior recognition. In this work, we investigate two mask estimation approaches, namely state dependent and deep neural network (DNN) based estimations, separate from noises for improving DNN acoustic models' noise robustness. second approach experimentally shown outperform first one. Due stereo...
Purpose – The purpose of this paper is to present a novel method for fabric defect detection. Design/methodology/approach based on joint low-rank and spare matrix recovery, since patterned manufactured by set predefined symmetry rules, it can be seen as the superposition sparse defective regions defect-free regions. A robust principal component analysis model with noise term designed handle images diverse patterns robustly. authors also estimate prior use guide recovery process accurate...
We consider the problem of target estimation in distributed MIMO radars that employ compressive sensing. The is formulated as a sparse signal recovery with magnitude constraints on reflection coefficients, where to be recovered consist equal size blocks have same sparsity profile. A solution proposed based alternating direction method multipliers (ADMM), which significantly lowers computational complexity and improves accuracy. Due block diagonal structure sensing matrix, iterations all ADMM...
Text-driven image editing aims to manipulate images with the guidance of natural language description. Text is much more and intuitive than many other interaction modes, attracts attention recently. However, compared classical supervised learning tasks, there no standard benchmark dataset for text-driven interactive up now. Therefore, it hard train an end-to-end model pixel-aligned driven by text. Some methods follow paradigm text-to-image models incorporating target into process generation....
Most of the existing 3D talking face synthesis methods suffer from lack detailed facial expressions and realistic head poses, resulting in unsatisfactory experiences for users. In this paper, we propose a novel pose-aware method with geometry-guided audio-vertices attention. To capture more expression, such as subtle nuances mouth shape eye movement, to build hierarchical audio features including global attribute feature series vertex-wise local latent movement features. Then, order fully...
State-of-the-art inter-datacenter WANs rely on software defined networking (SDN) to orchestrate their data transmission. Optimization requires frequent network update operations switch forwarding tables. When scheduling WANs, the utility of services should be respected. Yet, existing approaches do not respect and could result in performance degradation during procedure. Further, causes only degradation, but also period is unnecessarily prolonged. In this paper we propose Hermes, a...