A5102714720- Remote Sensing in Agriculture
- Land Use and Ecosystem Services
- Marine and coastal ecosystems
- Remote-Sensing Image Classification
- Advanced Image and Video Retrieval Techniques
- Topic Modeling
- Reinforcement Learning in Robotics
- Hydrology and Watershed Management Studies
- Impact of Light on Environment and Health
- Flood Risk Assessment and Management
- Vehicle emissions and performance
- Robotics and Sensor-Based Localization
- COVID-19 impact on air quality
- Medical Image Segmentation Techniques
- Advanced Image Fusion Techniques
- Calibration and Measurement Techniques
- Text and Document Classification Technologies
- Tropical and Extratropical Cyclones Research
- Environmental Changes in China
- Air Quality Monitoring and Forecasting
- Multimodal Machine Learning Applications
- Landslides and related hazards
- Urban Heat Island Mitigation
- Remote Sensing and Land Use
- Atmospheric aerosols and clouds
Nanjing University
2023-2024
Chang'an University
2020-2023
Xi’an University
2022-2023
Northwestern Polytechnical University
2021-2023
Jangan University
2020-2021
University Research Co (United States)
2021
Beijing Institute of Technology
2017
We formulate the registration as a function that maps input reference and sensed images to eight displacement parameters between prescribed matching points, opposed usual techniques (feature extraction–description–matching–geometric restrictions). The projection transformation matrix (PTM) is then computed in neural network used warp image, uniting all tasks under one framework. In this article, we offer multimodal image fusion with self-attention merge feature representation of images....
Current Reinforcement Learning (RL) methods often suffer from sample-inefficiency, resulting blind exploration strategies that neglect causal relationships among states, actions, and rewards. Although recent approaches aim to address this problem, they lack grounded modeling of reward-guided understanding states actions for goal-orientation, thus impairing learning efficiency. To tackle issue, we propose a novel method named Causal Information Prioritization (CIP) improves sample efficiency...
With the rapid advancement of Large Language Models (LLMs), safety LLMs has been a critical concern requiring precise assessment. Current benchmarks primarily concentrate on single-turn dialogues or single jailbreak attack method to assess safety. Additionally, these have not taken into account LLM's capability identifying and handling unsafe information in detail. To address issues, we propose fine-grained benchmark SafeDialBench for evaluating across various attacks multi-turn dialogues....
We propose a novel remote sensing image registration method based on the deep learning regression network. Different from traditional methods of feature extraction and matching, we pair blocks sensed reference images, then directly learn displacement parameters four corners block relative to image. In addition, develop dual network with weight sharing fully extract features. The proposed is tested different period Landsat-7 WorldView-3 images compared scale-invariant transform (SIFT), fast...
The distribution and characteristics of geological lineaments in areas with active faulting are vital for providing a basis regional tectonic identification analyzing the significance. Here, we extracted Qianhe Graben, an mountainous area on southwest margin Ordos Block, China, by using tensor voting algorithm after comparing them segment tracing (STA) LINE PCI Geomatica Software. main results show that (1) this mostly induced fault events trending NW–SE, (2) box dimensions all lineaments,...
Abstract In this paper, Yuqiao Reservoir is taken as the research object. The total suspended matter (TSM) produced by economic development in upper reaches of reservoir and its surrounding areas has brought great ecological harm to safe operation reservoir. Satellite remote sensing technology provides a good way obtain temporal spatial variation TSM study area. Two field surveys were carried out Reservoir, 44 sampling points collected two tests. spectral data concentration obtained. We...
Currently available high-resolution digital elevation model (DEM) is not particularly useful to geologists for understanding the long-term changes in fluvial landforms induced by tectonic uplift, although DEMs that are generated from satellite stereo images such as ZiYuan-3 (ZY3) include characteristics with significant coverage and rapid acquisition. Since an ongoing analysis of systems lacking, ZY3 DEM was block adjustment describe mountainous area Qianhe Basin have been uplift. Moreover,...
中国城市黑臭水体情况严重,基于遥感监测黑臭水体刚刚起步,很多问题待解决。以沈阳市城市建成区内主要河流为研究区,于2015年—2016年开展地面调查,获取了浑河和蒲河46个一般水体的样点,和辉山明渠、满堂河、细河以及微山湖路附近、丁香湖北部50个黑臭水体的样点数据,包括水面光谱和主要水质参数。分析了黑臭水体与一般水体的光谱特征,发现城市黑臭水体反射率光谱在绿光—红光波段变化比一般水体平缓,基于这一特点提出了一种基于反射率光谱指数BOI(Black and Odorous water...
As an air pollutant closely related to urban traffic and heavy industrial capacity, the variation of NO2 (nitrogen dioxide) concentration can directly reflect strength socioeconomic activities. Using weekly average results daily product synthesis tropospheric column concentrations from OMI (Ozone Monitoring Instrument) satellite inversion, a weekly-scale series standardized activity index during Spring Festival period 2019–2021 is constructed. The show that OMI-NO2 data are in good...
We propose a semantic mapping-based remote sensing image matching method, which aims to obtain the positions of candidate patches containing keypoints directly on reference image, avoiding use cost-volume search pixel by pixel. First, global context-fusing attention structure is created fuse information for with entire sensed image. Then, self-attention layer dependencies proposed extract cross-modal representation. The receptive field provided enables method mapping experimental results...
Advances in the object-based convolutional neural network (CNN) have demonstrated superiority of CNNs for image classification. However, any CNN, regardless its model structure, only stacks square images with different scales when extracting features. The impact background information around segmented object (the number pixels object) classification accuracy is neglected. In addition, blurred boundaries and feature representation, as well huge computational redundancy, restrict application...
To achieve a better robustness and get static response of certain target missiles in longitudinal control, mathematical model is established. The uncontrolled trajectory, the predetermined trajectory requirements are analyzed next. Then instruction angle designed to ensure stable flight system. solve problem that PID controller cannot adjust parameters online output easy be disturbed by external noise, fuzzy adaptive control designed. It enhanced also has good using control. simulation...
Purpose Online question-and-answer (Q&A) communities serve as important channels for knowledge diffusion. The purpose of this study is to investigate the dynamic development process online systems and explore final or progressive state system development. By measuring nonlinear characteristics from perspective complexity science, authors aim enrich method research on dynamics systems, deeply understand behavior rules systems. Design/methodology/approach collected data programming-related...
Self-learning and social learning stand as two pivotal constituents in multi-agent exploration. Inspired by the fact that animals humans explore unfamiliar environments to learn survival skills training themselves using unlabeled data replicating others' successful experiences, we propose a reinforcement method, named Self-Learning Social Learning (S <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> L), which aims address complex tasks...
Most real-world multi-agent tasks exhibit the characteristic of sparse interaction, wherein agents interact with each other in a limited number crucial states while largely acting independently. Effectively modeling interaction and leveraging learned structure to instruct agents' learning processes can enhance efficiency reinforcement algorithms. However, it remains unclear how identify these specific interactive solely through trials errors within current tasks. To address this challenge,...
Abstract The accuracy of atmospheric correction is a key factor affecting the quantitative application GF-2 satellite. Due to lack short-wave infrared band in GF-2, it impossible use method dark pixel for correction. Therefore, coefficient lookup table (LUT) based on 6S (Second Simulation Satellite Signal Solar Spectrum) radiation transfer model and aerosol optical thickness (AOT) parameters retrieved from MODIS data proposed, satellite multispectral carried out. Dunhuang field with flat...