A5081433919- Image Enhancement Techniques
- Advanced Image Fusion Techniques
- Hydrology and Watershed Management Studies
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Older Adults Driving Studies
- Aquatic Ecosystems and Phytoplankton Dynamics
- Advanced Neural Network Applications
- AI in cancer detection
- Soil and Water Nutrient Dynamics
- Traffic and Road Safety
- Brain Tumor Detection and Classification
- Medical Image Segmentation Techniques
- Image and Video Quality Assessment
- Color Science and Applications
- Video Surveillance and Tracking Methods
- Water Quality Monitoring and Analysis
- Water resources management and optimization
- Water Quality and Pollution Assessment
- Anomaly Detection Techniques and Applications
- Human Pose and Action Recognition
- Environmental Changes in China
- Environmental and Agricultural Sciences
- Advanced Photocatalysis Techniques
- Copper-based nanomaterials and applications
Chinese Research Academy of Environmental Sciences
2011-2025
Sun Yat-sen University
2024
China University of Political Science and Law
2024
Beijing Institute of Radio Metrology and Measurement
2024
Massachusetts General Hospital
2020-2023
Chinese Academy of Medical Sciences & Peking Union Medical College
2023
Harvard University
2016-2022
Sichuan University of Science and Engineering
2022
Hefei University of Technology
2021
Anhui Special Equipment Inspection Institute
2021
Image enhancement plays an important role in image processing and analysis. Among various algorithms, Retinex-based algorithms can efficiently enhance details have been widely adopted. Since regard illumination removal as a default preference fail to limit the range of reflectance, naturalness non-uniform images cannot be effectively preserved. However, is essential for achieve pleasing perceptual quality. In order preserve while enhancing details, we propose algorithm images. general, this...
Enhancement of non-uniformly illuminated images often suffers from over-enhancement and produces unnatural results. This paper presents a naturalness preserved enhancement method for images, using priori multi-layer lightness statistics acquired high-quality images. Our work makes three important contributions: designing novel image model; deriving the outdoor which are incorporated into showing that overall quality rating enhanced is consistent with combination contrast preservation. Two...
As a challenging problem, image haze removal plays an important role in computer vision applications. The dark channel prior has been widely studied for since it is simple and effective; however, still suffers from over‐saturation, artefacts dark‐look. To resolve these problems, this study proposes method of single using content‐adaptive post enhancement. main contributions work are as follows: first, associative filter, which can transfer the structures reference grey levels coarse to...
The Sanjiang Plain (SJP) in Northeast China, a crucial black soil region, serves as quintessential example of high-intensity agricultural development zone and stands China’s largest commercial grain production base. In the context global climate change, pronounced warming increased vegetation greening are expected to significantly impact water resource supply its alignment with crop requirements SJP. This study assesses how change affect supply–demand relationship SJP, addressing critical...
The nitrogen cycle, essential for the transformation and circulation of on earth, involves key processes such as fixation, nitrification, denitrification, ammonification. Microorganisms are crucial in these processes, influencing water quality through energy conversion cycling. Eutrophication, driven by human activities, increases exogenous input, accelerating cycling loss, boosting greenhouse gas emissions, impacting aquatic ecosystems. Functional genes amoA (nitrification), nirS, nirK,...
In this paper, an efficient image enhancement method based on Retinex and lightness decomposition is proposed, which enhances details preserves the naturalness simultaneously. The quality of enhanced determined by two factors, naturalness. Accordingly, proposed to be decomposed into reflex ambience illumination. utilized extract Retinex, illumination maintain scenes. Moreover, a coarse evaluation that keeps corresponding order substitute absolute amount illumination, may not necessary...
Outdoor images captured in bad-weather conditions usually have poor intensity contrast and color saturation since the light arriving at camera is severely scattered or attenuated. The task of improving image quality remains a challenge. Existing methods improvement are effective for small group but often fail to produce satisfactory results broader variety images. In this paper, we propose an enhancement method, which makes it applicable enhance outdoor by using content-adaptive as well...
This paper proposes a novel U-Net variant using stacked dilated convolutions for medical image segmentation (SDU-Net). SDU-Net adopts the architecture of vanilla with modifications in encoder and decoder operations (an operation indicates all processing feature maps same resolution). Unlike which incorporates two standard each encoder/decoder operation, uses one convolution followed by multiple concatenates outputs as input to next operation. Experiments showed that outperformed U-Net,...
Information about the growth, productivity, and distribution of vegetation, which are highly relied on sensitive to natural anthropogenic factors, is essential for agricultural production management eco-environmental sustainability in Amur River Basin (ARB). In this paper, spatial–temporal trends vegetation dynamics were analyzed at pixel scale ARB period 1982–2013 using remotely sensed data long-term leaf area index (LAI), fractional cover (FVC), terrestrial gross primary productivity...
Some drugs can be used to treat multiple diseases, suggesting potential patterns in drug treatment. Determination of treatment improve our understanding the mechanisms action, enabling repurposing. A associated with a multilayer tissue-specific protein–protein interaction (TSPPI) network for diseases it is treat. Proteins usually interact other proteins achieve functions that cause diseases. Hence, studying similar common module structures TSPPI networks. Therefore, we propose network-based...
The Sanjiang Plain (SJP), located at the confluence reaches of Heilong, Songhua, and Wusuli Rivers in Northeast China. This study aimed to quantify effects varying climate land-use/land-cover (LULC) dynamics on green water (GW) over SJP during two distinctive periods (i.e., pre-2000 post-2000), when synergetic increased precipitation temperature rapid development agriculture occurred. assessment used distributed eco-hydrological model ESSI-3. Multivariable multi-objective calibration...
Analyzing naturalistic driving behavior recorded with in-car cameras is an ecologically valid method for measuring errors, but it time intensive and not easily applied on a large scale. This study validated semi-automated, computerized using archival data collected drivers mild Alzheimer’s disease (AD; n = 44) age-matched healthy controls (HC; 16). The flagged situations where safety concerns are most likely to occur (i.e., rapid stops, lane deviations, turns, intersections). These epochs...
Purpose: The purpose of this study was to investigate the telescope use behaviors in natural daily driving people with reduced visual acuity licensed drive a bioptic (a small spectacle-mounted telescope). Methods: A large dataset (477 hours) naturalistic collected from 19 drivers (visual 20/60 20/160 without To reduce data volume, multiloss 50-layer deep residual neural network (ResNet-50) used detect potential events. Then, total 120 hours selected video clips were reviewed and annotated...
Gaze-contingent displays have been widely used in vision research and virtual reality applications. Due to data transmission, image processing, display preparation, the time delay between eye tracker monitor update may lead a misalignment position manipulation during movements. We propose method reduce using Taylor series predict saccadic movement. The proposed was evaluated two large datasets including 219,335 human saccades (collected with an EyeLink 1000 system, 95% range from 1° 32°)...