A5100682003- Arctic and Antarctic ice dynamics
- Methane Hydrates and Related Phenomena
- Ocean Waves and Remote Sensing
- Underwater Acoustics Research
- Oceanographic and Atmospheric Processes
- Climate change and permafrost
- Cryospheric studies and observations
- Marine Biology and Ecology Research
- Robotics and Sensor-Based Localization
- Precipitation Measurement and Analysis
- Geological and Geochemical Analysis
- earthquake and tectonic studies
- Advanced Vision and Imaging
- Topic Modeling
- Geochemistry and Geologic Mapping
- Marine Bivalve and Aquaculture Studies
- Image Enhancement Techniques
- Anomaly Detection Techniques and Applications
- Soil Moisture and Remote Sensing
- Network Security and Intrusion Detection
- Ship Hydrodynamics and Maneuverability
- Marine and coastal plant biology
- Wireless Signal Modulation Classification
- Tropical and Extratropical Cyclones Research
- Robotic Path Planning Algorithms
South China University of Technology
2024-2025
Minzu University of China
2023-2025
Tsinghua University
2023-2025
Minjiang University
2013-2024
University of Waterloo
2022-2024
Chengdu University of Technology
2022-2024
State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation
2022-2024
University of Illinois Urbana-Champaign
2024
University of Pennsylvania
2019-2024
Memorial University of Newfoundland
2018-2022
Since rain alters the histogram pattern of radar images, rain-contaminated data can be identified. In this article, a support vector machine (SVM)-based method for detection using X-band marine images is presented. First, normalized bin values each image are extracted and combined into feature vector. Then, SVMs employed to classify between rain-free images. Radar simultaneous rate collected from sea trial in North Atlantic Ocean utilized model training testing. Comparison with zero pixel...
Bayesian neural networks (BNNs) have been demonstrated to be effective in accurate retrieval of sea ice concentration (SIC) from multi-source data, while providing estimates uncertainty, which are essential for downstream services. However, uncertainty obtained by BNNs intrinsically uncalibrated, indicates that it may not correlate well with model error. To address this issue, we investigate a new approach combines an auxiliary prediction interval (PI) estimator the BNN-based SIC mean...
Sea ice mapping on synthetic aperture radar (SAR) imagery is important for various purposes, including ship navigation and usage in environmental climatological studies. Although a series of deep learning-based models have been proposed automatic sea classification SAR scenes, most them are flat N-way classifiers that do not consider the uneven visual separability different types. To further improve accuracy with limited training samples, hierarchical pipeline from SAR. First, semantic...
In this article, an unsupervised clustering-based method for identifying rain-contaminated and low-backscatter regions in X-band marine radar images is presented. Rain blurs the wave signatures of images, caused by calibration errors or too-low wind speed contain little no signatures. both cases, ocean surface parameter measurement using will be negatively affected. Four types features can extracted based on distinct difference texture pixel intensity distribution between rain-free,...
Spatial–temporal features are extracted from X-band marine radar backscatter image sequences via deep neural networks to estimate sea surface significant wave heights (SWHs). A convolutional network (CNN) is first constructed based on the pretrained GoogLeNet SWH using multiscale spatial each image. Since CNN-based model cannot analyze temporal behavior of signatures in sequences, a gated recurrent unit (GRU) concatenated after layers CNN build GRU (CGRU)-based model, which generates...
In recent years, the adoption of deep learning (DL) techniques for predicting sea ice concentration (SIC) given both passive microwave (PM) data and reanalysis has seen a growing interest. For use in downstream services, these estimates should be accompanied by uncertainty estimates. To provide estimates, we utilize heteroscedastic Bayesian neural network (HBNN), which can estimate model (epistemic) (aleatoric) uncertainty. We PM atmospheric as our input features, demonstrate that are needed...
Algorithms designed for ice–water classification of synthetic aperture radar (SAR) sea ice imagery produce only binary (ice and water) output typically using manually labeled samples assessment. This is limiting because a small subset are used, which, given the nonstationary nature water classes, will likely not reflect full scene. To address this, we implement in more informative manner considering uncertainty associated with each pixel accomplish have implemented Bayesian convolutional...
The southwestern Yangtze Block represents one of the world's largest lead–zinc provinces, comprises more than 400 known carbonate-hosted Pb-Zn deposits and contains over 30 million tons (Mt) total Pb + Zn reserves. in this area have been studied extensively by economic geologists, but origin mineralization processes these are still controversial. Yuanbaoshan deposit is a typical medium scale (0.19 Mt P reserves @ 0.7–26.31 wt%), which northern part mineral belt southwest margin Block. In...
Abstract. The AutoICE challenge, organized by multiple national and international agencies, seeks to advance the development of near-real-time sea ice products with improved spatial resolution, broader temporal coverage, enhanced consistency. In this paper, we present a detailed description our solutions experimental results for challenge. We have implemented an automated mapping pipeline based on multi-task U-Net architecture, capable predicting concentration (SIC), stage (SOD), floe size...
Abstract. Mapping sea ice in the Arctic is essential for maritime navigation, and growing vessel traffic highlights necessity of timeliness accuracy charts. In addition, with increased availability satellite imagery, automation becoming more important. The AutoICE Challenge investigates possibility creating deep learning models capable mapping multiple parameters automatically from spaceborne synthetic aperture radar (SAR) imagery assesses current state automatic-sea-ice-mapping scientific...
In this paper, a temporal convolutional network (TCN)-based model is proposed to retrieve significant wave height (Hs) from X-band nautical radar images. Three types of features are first extracted image sequences based on signal noise ratio (SNR), ensemble empirical mode decomposition (EEMD), and gray level co-occurrence matrix (GLCM) methods, respectively. Then, feature vectors input into the TCN-based regression produce Hs estimation. Radar data collected moving vessel at East Coast...
This study aims to address the problem of poor stem node detection and recognition accuracy in high-resolution global sugarcane images, which is crucial for accurately identifying locations during automated cutting. Since occupies a very small area entire image, only about 0.2%, feature blurring caused by scaling original image fit network input size requirements often results loss important information. In order overcome this problem, conducts an in-depth on target images proposes improved...
In the tandem turbine-based combined cycle engine's hyperburner, complex intrusive design is usually employed to achieve ignition and stable combustion under high-speed low-temperature extreme conditions. This paper proposes a concise, non-intrusive coupled swirler-cavity configuration supply fuel with good atomization at cavity-trapped vortex center, thereby ensuring formation of initial flame kernel. Under guidance vortex, kernel grows through dynamic “rotation-diffusion” mechanism....
Numerous intrusive bodies of mafic–ultramafic to felsic compositions are exposed in association with volcanic rocks the Late Permian Emeishan large igneous province (ELIP), southwestern China. Most granitic ELIP were derived by differentiation basaltic magmas a mantle connection, and crustal have rarely been studied. Here we investigate suite mafic dykes I-type granites that yield zircon U-Pb emplacement ages 259.9 ± 1.2 Ma 259.3 1.3 Ma, respectively. The εHf(t) values from DZ dyke –0.3 9.4,...
The presence of rain may blur surface wave signatures and cause additional radar backscatter, which negatively affects the performance ocean remote sensing applications (e.g., wind parameter measurement) using X-band marine radars. In this article, a novel end-to-end model is developed to detect locate rain-contaminated pixels in images based on type deep neural network called SegNet, able segment regions by classifying each pixel into three classes: rain-free, rain-contaminated,...
In this article, the accuracy of wave direction and period estimation from <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">X</i> -band marine radar images under different rain rates is analyzed, a simple subimage selection scheme proposed to mitigate effect. First, each image divided into multiple subimages, subimages with relatively clear signatures are identified based on random-forest-based classification model. Then, estimated by...
The presence of rain can negatively affect the performance many sensors such as X-band radar. In this paper, an effective approach is proposed to mitigate effect on significant wave height ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${H}_{s}$ </tex-math></inline-formula> ) estimation from radar sensor data along with a machine-learning (ML)-based method. First all, haze removal algorithm applied...
The presence of rain degrades the performance sea surface parameter estimation using X-band marine radar. In this article, a novel scheme is proposed to improve wind measurement accuracy from rain-contaminated radar data. After extracting texture features each image pixel, regions with blurry wave signatures are first identified self-organizing map (SOM)-based clustering model. Then, convolutional neural network used for haze removal, i.e., DehazeNet introduced and incorporated into...