- Remote-Sensing Image Classification
- Image Retrieval and Classification Techniques
- Remote Sensing and Land Use
- Advanced Image and Video Retrieval Techniques
- Image and Signal Denoising Methods
- Welding Techniques and Residual Stresses
- Remote Sensing and LiDAR Applications
- Bayesian Methods and Mixture Models
- Digital Media and Visual Art
- Automated Road and Building Extraction
- Remote Sensing in Agriculture
- Metallurgical Processes and Thermodynamics
- Medical Image Segmentation Techniques
- Air Quality Monitoring and Forecasting
- Satellite Image Processing and Photogrammetry
- Advanced Image Processing Techniques
- Climate Change and Health Impacts
- Advanced Welding Techniques Analysis
- Image Processing Techniques and Applications
- Landslides and related hazards
- Structural Engineering and Vibration Analysis
- Air Quality and Health Impacts
- Spectroscopy and Chemometric Analyses
- Microstructure and Mechanical Properties of Steels
- Blind Source Separation Techniques
State Grid Corporation of China (China)
2024
Montanuniversität Leoben
2024
State Key Laboratory of Remote Sensing Science
2018-2023
Beijing Normal University
2013-2023
Institute of Remote Sensing and Digital Earth
2018-2022
Chinese Academy of Sciences
2007-2022
University of Surrey
2002-2021
Hubei Urban Construction Vocational and Technological College
2021
Sichuan University
2021
West China Hospital of Sichuan University
2021
Higher resolution building mapping from lower remote sensing images is in great demand due to the lack of higher data access, especially context disaster assessment. High layout map crucial for emergency rescue after disaster. The response time would be reduced if detailed footprints were delineated more easily available low-resolution data. To achieve this goal, we propose a super-resolution semantic segmentation network called ESPC_NASUnet, which consists feature module and module. best...
Recently, several recurrent neural network (RNN)-based models have been proposed to delineate the outlines of buildings from very high resolution (VHR) remote sensing images. These first use convolutional networks (CNNs) recognize boundary fragments by learning probability maps both edges and corners then feed them into RNN find link a set sequent external boundaries buildings. However, caused category imbalance corners, local ambiguity edge detection is serious, which significantly affects...
Abstract M-type asteroids are historically thought to be exposed metallic cores of differentiated planets with a composition dominated by pure iron and nickel. However, recent spectral radar observations reveal an insufficient number in the main belt. Here, we report unusual space weathering characteristics associated natural metal grain found Chang'e 5 lunar soil. Microcraters, impact glass, whiskers, unique vesicular rims on surface this help explain properties some potential asteroids,...
The demand for automatically recognizing and retrieving medical images screening, reference, management is growing faster than ever. In this paper, we present an intelligent content-based image retrieval system called I-Browse, which integrates both iconic semantic content histological analysis. I-Browse combines low-level processing technology with high-level analysis of through different modules in the proposed architecture. Similarity measures are their performance evaluated. Furthermore,...
It is an important task to automatically and accurately map rooftops from very high resolution remote sensing images since buildings are closely related human activity. Two typical technologies often utilized accomplish the task, i.e., semantic segmentation instance segmentation. The independently allocate a label (e.g., “building” or not) each pixel, resulting in blob-like segments. On contrary, one might model boundary of rooftop as polygon improve shape by encouraging vertices adhere...
Deep learning-based semantic segmentation has made great progress in understanding very-high-resolution (VHR) remote sensing images (RSIs). However, large-scale applications are still limited. The main reason is that diverse imaging modes and geographical differences make it difficult to transfer a model trained the source domain target domain. To solve this problem, unsupervised adaptation (UDA) for VHR RSIs received some attention, but accuracy of cross-domain needs be improved. Currently,...
In this letter, we present a novel object-oriented semantic clustering algorithm for high-spatial-resolution remote sensing images using the probabilistic latent analysis (PLSA) model coupled with neighborhood spatial information. First of all, an image collection is generated by partitioning large satellite into densely overlapped subimages. Then, PLSA employed to collection. Specifically, partitioned two subsets. One used learn topic models, where number topics determined minimum...
Real-time estimation of crop progress stages is critical to the US agricultural economy and decision making. In this paper, a Hidden Markov Model (HMM) based method combining multisource features has been presented. The include mean Normalized Difference Vegetation Index (NDVI), fractal dimension, Accumulated Growing Degree Days (AGDDs). our case, these are global variable, measured in state-level. Moreover, feature each Day Year (DOY) would be impacted by multiple stages. Therefore, mixture...
Detailed information about built-up areas is valuable for mapping complex urban environments. Although a large number of classification algorithms such have been developed, they are rarely tested from the perspective feature engineering and learning. Therefore, we launched unique investigation to provide full test Operational Land Imager (OLI) imagery 15-m resolution area in 2015, Beijing, China. Training classifier requires many sample points, proposed method based on European Space...
Automated audio captioning aims to use natural language describe the content of data. This paper presents an system with encoder-decoder architecture, where decoder predicts words based on features extracted by encoder. To improve proposed system, transfer learning from either upstream audio-related task or a large in-domain dataset is introduced mitigate problem induced data scarcity. Besides, evaluation metrics are incorporated into optimization model reinforcement learning, which helps...
Since buildings are closely related to human activities, large-scale mapping of individual has become a hot research topic. High-resolution images with sub-meter or meter resolution common choices produce maps building footprints. However, high-resolution both infrequently collected and expensive obtain process, making it very difficult timely. This paper presents simple but effective way national-scale map footprints using feature super-resolution semantic segmentation sentinel-2 images....
The dynamics of weld solidification were observed in situ using a laser welding process on transparent organic materials systems. Succinonitrile was used to simulate pure metal system and succinonitrile with 1.2 wt. % acetone an alloy system. Observed pool shapes good agreement theoretical heat transfer calculations. shape the succinonitrile-acetone related complex interactions between grain orientation, selection, dendrite orientations, which depend strongly speed. An increase speed leads...
Two-step ways are often used for fusing both panchromatic (PAN) and multispectral (MS) images classification, e.g., classifying MS sharpened by PAN or directly pouring fine spatial details of into a classification result images. In this paper, we present unified Bayesian framework to iteratively discovering semantic segments from allocating cluster labels the using Specifically, probabilistic generative process is explained with generalized metaphor Chinese restaurant franchise (CRF) (gCRF),...
Punctuation is critical in understanding natural language text.Currently, most automatic speech recognition (ASR) systems do not generate punctuation, which affects the performance of downstream tasks, such as intent detection and slot filling.This gives rise to need for punctuation restoration.Recent work restoration heavily utilizes pre-trained models without considering data imbalance when predicting classes.In this work, we address problem by proposing a token-level supervised...
Automated Audio captioning (AAC) is a cross-modal translation task that aims to use natural language describe the content of an audio clip. As shown in submissions received for Task 6 DCASE 2021 Challenges, this problem has increasing interest community. The existing AAC systems are usually based on encoder-decoder architecture, where signal encoded into latent representation, and aligned with its corresponding text descriptions, then decoder used generate captions. However, training system...
Abstract Immunocompromised hosts often die of acute infection, presumably, due to lack an adaptive immune response clear pathogens. However, we have now demonstrated that unleashed innate can also be a direct cause death. After viral nude mice produced higher levels proinflammatory cytokines but died even though titers virus remained comparable wild type mice. Administration poly I:C, ligand for toll-like receptor 3, led cytokine storm in immunocompromised NK cell and TNF dependent fashion....
Some basic issues on content-based remote sensing image retrieval are discussed in this paper. The framework, processing flow and levels proposed based theory of CBIR characteristics RS image. Oriented to the practical demands, five patterns including template-based, attribute-based, metadata-based, semanteme-based integrated proposed. contents features that can be used include color, shape, texture, spectra, spatial relation, metadata relative rules knowledge. Among those features, spectral...