A5002166815- Image Enhancement Techniques
- Advanced Vision and Imaging
- Smart Grid Energy Management
- Power System Optimization and Stability
- Optimal Power Flow Distribution
- Video Surveillance and Tracking Methods
- Advanced Image Processing Techniques
- Microgrid Control and Optimization
- Energy Load and Power Forecasting
- Visual Attention and Saliency Detection
- Smart Grid Security and Resilience
- Image Processing Techniques and Applications
- Face and Expression Recognition
- Power System Reliability and Maintenance
- Power Systems and Technologies
- Image and Object Detection Techniques
- Integrated Energy Systems Optimization
- Text and Document Classification Technologies
- Industrial Vision Systems and Defect Detection
- Power Systems Fault Detection
- Advanced Neural Network Applications
- Advanced Measurement and Detection Methods
- Electric Power System Optimization
- Advanced Image Fusion Techniques
- Optical measurement and interference techniques
Hunan Normal University
2016-2025
Shihezi University
2025
First Affiliated Hospital of Jinan University
2025
Zhengzhou University
2024
Hunan University
2009-2024
Intel (United States)
2023-2024
Liaoning Technical University
2023
Changsha Normal University
2018-2022
Xinjiang Medical University
2022
First Affiliated Hospital of Xinjiang Medical University
2022
In this paper, a reinforcement-learning-based online optimal (RL-OPT) control method is proposed for the hybrid energy storage system (HESS) in ac-dc microgrids involving photovoltaic systems and diesel generators (DGs). Due to low inertia, conventional unregulated charging discharging (C&D) of storages may introduce disturbances that degrade power quality performance, especially fast C&D situations. Secondary tertiary levels can optimize state charge reference HESS; however, they are...
Glass bottles must be thoroughly inspected before they are used for packaging. However, the vision inspection of bottle bottoms defects remains a challenging task in quality control due to inaccurate localization, difficulty detecting texture region, and intrinsically nonuniform brightness across central panel. To overcome these problems, we propose surface defect detection framework, which is composed three main parts. First, new localization method named entropy rate superpixel circle...
The concept of peer-to-peer (P2P) trading, or transactive energy (TE), is gaining momentum as a future grid restructure. It has the potentials to utilize distributed resources (DERs), proactive demand side management (DSM), and infusion in information communication technologies (e.g., blockchain Internet Things (IoT)) for promoting technical economic efficiency system its entirety. An efficient market framework vital successful sustainable implementation such concept. This article proposes...
Glass bottles are widely used as containers in the food and beverage industry, especially for beer carbonated beverages. As key part of a glass bottle, bottle bottom its quality closely related to product safety. Therefore, must be inspected before is packaging. In this paper, an apparatus based on machine vision designed real-time inspection, framework defect detection mainly using saliency template matching presented. Following brief description apparatus, our emphasis image analysis....
Learning a robust object detector in adverse weather with real-time efficiency is of great importance for the visual perception task autonomous driving systems. In this article, we propose framework to improve YOLO detector, denoted as R(obust)-YOLO, without need annotations weather. Considering distribution gap between normal images and images, our consists an image quasi-translation network (QTNet) feature calibration (FCNet) adapting domain gradually. Specifically, use simple yet...
Previous methods in salient object detection (SOD) mainly focused on favorable illumination circumstances while neglecting the performance low-light condition, which significantly impedes development of related down-stream tasks. In this work, considering that it is impractical to annotate large-scale labels for task, we present a framework (HDNet) detect objects images with synthetic images. Our HDNet consists foreground highlight sub-network (HNet) and an appearance-aware (DNet), both can...
Histogram equalisation has been widely used for image enhancement because of its simple implementation and satisfactory performance. However, traditional histogram uniformly redistributes an entire or multiple piecewise histograms with the same strategy, which may produce unnatural artefacts, over‐enhancement under‐enhancement in wide dynamic range dark enhancement. This study proposes adaptive extended algorithm (AEPHE) First, original is divided into a group histograms. Then, equalisation,...
Domain adaptive semantic segmentation enables robust pixel- wise understanding in real-world driving scenes. Source-free domain adaptation, as a more practical technique, addresses the concerns of data privacy and storage limitations typical unsupervised adaptation methods, making it especially relevant context intelligent vehicles. It utilizes well-trained source model unlabeled target to achieve domain. However, absence labels, current solutions cannot sufficiently reduce impact shift...
Face detection technology has widely attracted attention due to its enormous application value and market potential, such as face recognition video surveillance system. Real-time not only is one part of the automatic system but also developing an independent research subject. So, there are many approaches solve detection. Here modified AdaBoost algorithm based on OpenCV presented, experiments real-time detecting given through two methods timer dual-thread. The result shows that method with...
Short-term load forecasting (STLF) is essential for the reliable and economic operation of power systems. Though many STLF methods were proposed over past decades, most them focused on loads at high aggregation levels only. Thus, low-aggregation forecast still requires further research development. Compared with substation or city level loads, individual are typically more volatile much challenging to forecast. To address this issue, paper first discusses characteristics small-and-medium...
It is challenging to annotate large-scale datasets for supervised video shadow detection methods. Using a model trained on labeled images the frames directly may lead high generalization error and temporal inconsistent results. In this paper, we address these challenges by proposing Spatio-Temporal Interpolation Consistency Training (STICT) framework rationally feed unlabeled together with into an image network training. Specifically, propose Spatial Temporal ICT, in which define two new...
Unsupervised domain adaptation has attracted widespread attention as a promising method to solve the labeling difficulties of semantic segmentation tasks. It trains network for unlabeled real target images using easily available labeled virtual source images. To improve performance, clustering is used obtain domain-invariant feature representations. However, most clustering-based methods indiscriminately cluster all features mapped by category from both domains, causing centroid shift and...
Single image defogging algorithms based on prior assumptions or constraints have captured much attention because of their simplicity and practicality. However, they still some challenges to deal with foggy images under weather conditions where these may not be effective efficient enough. In this paper, we aim develop a novel algorithm by directly predicting the fog density recovered rather than adopting constraints. order achieve goal, two specific steps are introduced. First, adopt three...
State-of-the-art single image dehazing algorithms have some challenges to deal with images captured under complex weather conditions because their assumptions usually do not hold in those situations. In this paper, we develop a deep transmission network for robust dehazing. This simultaneously copes three color channels and local patch information automatically explore exploit haze-relevant features learning framework. We further different structures parameter settings achieve tradeoffs...
Traffic sign recognition is a rather challenging task for intelligent transportation systems since signs in different subsets, e.g., speed limit signs, prohibition and mandatory are very from each other color or shape, whereas they share some similarities to the ones same subset. Therefore, it important integrate modalities of visual features, such as select discriminative features better description; addition, benefits explore correlations between classes traffic learn classifiers jointly...
This study investigated the neuroanatomical basis of association between depression/anxiety and sleep quality among 370 college students. The results showed that there was a significant correlation depression/anxiety. Moreover, mediation gray matter volume right insula mediated relationship quality, which suggested may affect through volume. These findings confirmed strong link depression/anxiety, while highlighting volumetric variation in associated with emotional processing, play critical...
To explore the mediating role of sleep duration in relationship between depression symptoms and myopia among middle school students. This study was a cross-sectional research conducted using stratified cluster random sampling method. A total 1 728 students were selected from two junior high schools senior certain urban areas farms Xinjiang Production Construction Corps. Questionnaire surveys vision tests Spearman analysis used to analyze correlation symptoms, duration, myopia. The Bootstrap...