The problem of recognizing natural scenes, such as water, smoke, fire, wind-blown vegetation and a flock of flying birds, is considered. These scenes exhibit the characteristic dynamic pattern, but have stochastic ext...The problem of recognizing natural scenes, such as water, smoke, fire, wind-blown vegetation and a flock of flying birds, is considered. These scenes exhibit the characteristic dynamic pattern, but have stochastic extent. They are referred to as dynamic texture（DT）. In reality, the diversity of DTs on different viewpoints and scales are very common, which also bring great difficulty to recognize DTs. In the previous studies, due to no considering of the deformable and transient nature of elements in DT, the motion estimation method is based on brightness constancy assumption,which seem inappropriate for aggregate and complex motions. A novel motion model based on relative motion in the neighborhood of two-dimensional motion fields is proposed. The estimation of non-rigid motion of DTs is based on the continuity equation, and then the local vector difference（LVD） is proposed to characterize DT local relative motion. Spatiotemporal statistics of the LVDs is used as the representation of DT sequences. Excellent performances of classifying all DTs in UCLA database demonstrate the capability of the proposed method in describing DT.展开更多
Unsupervised learning methods such as graph contrastive learning have been used for dynamic graph represen-tation learning to eliminate the dependence of labels.However,existing studies neglect positional information ...Unsupervised learning methods such as graph contrastive learning have been used for dynamic graph represen-tation learning to eliminate the dependence of labels.However,existing studies neglect positional information when learning discrete snapshots,resulting in insufficient network topology learning.At the same time,due to the lack of appropriate data augmentation methods,it is difficult to capture the evolving patterns of the network effectively.To address the above problems,a position-aware and subgraph enhanced dynamic graph contrastive learning method is proposed for discrete-time dynamic graphs.Firstly,the global snapshot is built based on the historical snapshots to express the stable pattern of the dynamic graph,and the random walk is used to obtain the position representation by learning the positional information of the nodes.Secondly,a new data augmentation method is carried out from the perspectives of short-term changes and long-term stable structures of dynamic graphs.Specifically,subgraph sampling based on snapshots and global snapshots is used to obtain two structural augmentation views,and node structures and evolving patterns are learned by combining graph neural network,gated recurrent unit,and attention mechanism.Finally,the quality of node representation is improved by combining the contrastive learning between different structural augmentation views and between the two representations of structure and position.Experimental results on four real datasets show that the performance of the proposed method is better than the existing unsupervised methods,and it is more competitive than the supervised learning method under a semi-supervised setting.展开更多
With the development of computer graphics,the three-dimensional(3D)visualization brings new technological revolution to the traditional cartography.Therefore,the topographic 3D-map emerges to adapt to this technologic...With the development of computer graphics,the three-dimensional(3D)visualization brings new technological revolution to the traditional cartography.Therefore,the topographic 3D-map emerges to adapt to this technological revolution,and the applications of topographic 3D-map are spread rapidly to other relevant fields due to its incomparable advantage.The researches on digital map and the construction of map database offer strong technical support and abundant data source for this new technology,so the research and development of topographic 3D-map will receive greater concern.The basic data of the topographic 3D-map are rooted mainly in digital map and its basic model is derived from digital elevation model(DEM)and 3D-models of other DEM-based geographic features.In view of the potential enormous data and the complexity of geographic features,the dynamic representation of geographic information becomes the focus of the research of topographic 3D-map and also the prerequisite condition of 3D query and analysis.In addition to the equipment of hardware that are restraining,to a certain extent,the 3D representation,the data organization structure of geographic information will be the core problem of research on 3D-map.Level of detail(LOD),space partitioning,dynamic object loading(DOL)and object culling are core technologies of the dynamic 3D representation.The object-selection,attribute-query and model-editing are important functions and interaction tools for users with 3D-maps provided by topographic 3D-map system,all of which are based on the data structure of the 3D-model.This paper discusses the basic theories,concepts and cardinal principles of topographic 3D-map,expounds the basic way to organize the scene hierarchy of topographic 3D-map based on the node mechanism and studies the dynamic representation technologies of topographic 3D-map based on LOD,space partitioning,DOL and object culling.Moreover,such interactive operation functions are explored,in this paper,as spatial query,scene editing and management of topographic 3D-map.Finally,this paper describes briefly the applications of topographic 3D-map in its related fields.展开更多
Background modeling and subtraction is a fundamental problem in video analysis. Many algorithms have been developed to date, but there are still some challenges in complex environments, especially dynamic scenes in wh...Background modeling and subtraction is a fundamental problem in video analysis. Many algorithms have been developed to date, but there are still some challenges in complex environments, especially dynamic scenes in which backgrounds are themselves moving, such as rippling water and swaying trees. In this paper, a novel background modeling method is proposed for dynamic scenes by combining both tensor representation and swarm intelligence. We maintain several video patches, which are naturally represented as higher order tensors,to represent the patterns of background, and utilize tensor low-rank approximation to capture the dynamic nature. Furthermore, we introduce an ant colony algorithm to improve the performance. Experimental results show that the proposed method is robust and adaptive in dynamic environments, and moving objects can be perfectly separated from the complex dynamic background.展开更多
How our flexible behavior was supported by the dynamic(re)coding in fronto parietal was explored by the research group led by Prof.Chen Antao(陈安涛)from the Faculty of Psychology,Southwest University with the collabo...How our flexible behavior was supported by the dynamic(re)coding in fronto parietal was explored by the research group led by Prof.Chen Antao(陈安涛)from the Faculty of Psychology,Southwest University with the collaboration of Prof.Tobias Egner from Duke University.The study was supported by the National Natural Science Foundation of China,and was recently published in Journal of Neuroscience(2017,0935-17).展开更多
Two global experiments were carried out to investigate the effects of dynamic vegetation processes on numerical climate simulations from 1948 to 2008.The NCEP Global Forecast System(GFS)was coupled with a biophysical ...Two global experiments were carried out to investigate the effects of dynamic vegetation processes on numerical climate simulations from 1948 to 2008.The NCEP Global Forecast System(GFS)was coupled with a biophysical model,the Simplified Simple Biosphere Model(SSi B)version 2(GFS/SSi B2),and it was also coupled with a biophysical and dynamic vegetation model,SSi B version 4/Top-down Representation of Interactive Foliage and Flora Including Dynamics(TRIFFID)(GFS/SSi B4/TRIFFID).The effects of dynamic vegetation processes on the simulation of precipitation,near-surface temperature,and the surface energy budget were identified on monthly and annual scales by assessing the GFS/SSi B4/TRIFFID and GFS/SSi B2 results against the satellite-derived leaf area index(LAI)and albedo and the observed land surface temperature and precipitation.The results show that compared with the GFS/SSiB2 model,the temporal correlation coefficients between the globally averaged monthly simulated LAI and the Global Inventory Monitoring and Modeling System(GIMMS)/Global Land Surface Satellite(GLASS)LAI in the GFS/SSi B4/TRIFFID simulation increased from 0.31/0.29(SSiB2)to 0.47/0.46(SSiB4).The correlation coefficients between the simulated and observed monthly mean near-surface air temperature increased from 0.50(Africa),0.35(Southeast Asia),and 0.39(South America)to 0.56,0.41,and 0.44,respectively.The correlation coefficients between the simulated and observed monthly mean precipitation increased from 0.19(Africa),0.22(South Asia),and 0.22(East Asia)to 0.25,0.27,and 0.28,respectively.The greatest improvement occurred over arid and semiarid areas.The spatiotemporal variability and changes in vegetation and ground surface albedo modeled by the GFS with a dynamic vegetation model were more consistent with the observations.The dynamic vegetation processes contributed to the surface energy and water balance and in turn,improved the annual variations in the simulated regional temperature and precipitation.The dynamic vegetation processes had the greatest influence on the spatiotemporal changes in the latent heat flux.This study shows that dynamic vegetation processes in earth system models significantly improve simulations of the climate mean status.展开更多
基金supported by the National Natural Science Foundation of China(41504115)the Shaanxi Province Natural Science Foundation(2015JQ6223)+2 种基金the Foundation of Strengthening Police Science and Technology from Ministry of Public Security(2015GABJC50)the International Technology Cooperation Plan Project of Shaanxi Province(2015KW-0142015KW-013)
文摘The problem of recognizing natural scenes, such as water, smoke, fire, wind-blown vegetation and a flock of flying birds, is considered. These scenes exhibit the characteristic dynamic pattern, but have stochastic extent. They are referred to as dynamic texture（DT）. In reality, the diversity of DTs on different viewpoints and scales are very common, which also bring great difficulty to recognize DTs. In the previous studies, due to no considering of the deformable and transient nature of elements in DT, the motion estimation method is based on brightness constancy assumption,which seem inappropriate for aggregate and complex motions. A novel motion model based on relative motion in the neighborhood of two-dimensional motion fields is proposed. The estimation of non-rigid motion of DTs is based on the continuity equation, and then the local vector difference（LVD） is proposed to characterize DT local relative motion. Spatiotemporal statistics of the LVDs is used as the representation of DT sequences. Excellent performances of classifying all DTs in UCLA database demonstrate the capability of the proposed method in describing DT.
文摘Unsupervised learning methods such as graph contrastive learning have been used for dynamic graph represen-tation learning to eliminate the dependence of labels.However,existing studies neglect positional information when learning discrete snapshots,resulting in insufficient network topology learning.At the same time,due to the lack of appropriate data augmentation methods,it is difficult to capture the evolving patterns of the network effectively.To address the above problems,a position-aware and subgraph enhanced dynamic graph contrastive learning method is proposed for discrete-time dynamic graphs.Firstly,the global snapshot is built based on the historical snapshots to express the stable pattern of the dynamic graph,and the random walk is used to obtain the position representation by learning the positional information of the nodes.Secondly,a new data augmentation method is carried out from the perspectives of short-term changes and long-term stable structures of dynamic graphs.Specifically,subgraph sampling based on snapshots and global snapshots is used to obtain two structural augmentation views,and node structures and evolving patterns are learned by combining graph neural network,gated recurrent unit,and attention mechanism.Finally,the quality of node representation is improved by combining the contrastive learning between different structural augmentation views and between the two representations of structure and position.Experimental results on four real datasets show that the performance of the proposed method is better than the existing unsupervised methods,and it is more competitive than the supervised learning method under a semi-supervised setting.
文摘With the development of computer graphics,the three-dimensional(3D)visualization brings new technological revolution to the traditional cartography.Therefore,the topographic 3D-map emerges to adapt to this technological revolution,and the applications of topographic 3D-map are spread rapidly to other relevant fields due to its incomparable advantage.The researches on digital map and the construction of map database offer strong technical support and abundant data source for this new technology,so the research and development of topographic 3D-map will receive greater concern.The basic data of the topographic 3D-map are rooted mainly in digital map and its basic model is derived from digital elevation model(DEM)and 3D-models of other DEM-based geographic features.In view of the potential enormous data and the complexity of geographic features,the dynamic representation of geographic information becomes the focus of the research of topographic 3D-map and also the prerequisite condition of 3D query and analysis.In addition to the equipment of hardware that are restraining,to a certain extent,the 3D representation,the data organization structure of geographic information will be the core problem of research on 3D-map.Level of detail(LOD),space partitioning,dynamic object loading(DOL)and object culling are core technologies of the dynamic 3D representation.The object-selection,attribute-query and model-editing are important functions and interaction tools for users with 3D-maps provided by topographic 3D-map system,all of which are based on the data structure of the 3D-model.This paper discusses the basic theories,concepts and cardinal principles of topographic 3D-map,expounds the basic way to organize the scene hierarchy of topographic 3D-map based on the node mechanism and studies the dynamic representation technologies of topographic 3D-map based on LOD,space partitioning,DOL and object culling.Moreover,such interactive operation functions are explored,in this paper,as spatial query,scene editing and management of topographic 3D-map.Finally,this paper describes briefly the applications of topographic 3D-map in its related fields.
基金supported by National Natural Science Foundation of China (Grant Nos. 11301137 and 11371036)the National Science Foundation of Hebei Province of China (Grant No. A2014205100
文摘Background modeling and subtraction is a fundamental problem in video analysis. Many algorithms have been developed to date, but there are still some challenges in complex environments, especially dynamic scenes in which backgrounds are themselves moving, such as rippling water and swaying trees. In this paper, a novel background modeling method is proposed for dynamic scenes by combining both tensor representation and swarm intelligence. We maintain several video patches, which are naturally represented as higher order tensors,to represent the patterns of background, and utilize tensor low-rank approximation to capture the dynamic nature. Furthermore, we introduce an ant colony algorithm to improve the performance. Experimental results show that the proposed method is robust and adaptive in dynamic environments, and moving objects can be perfectly separated from the complex dynamic background.
文摘How our flexible behavior was supported by the dynamic(re)coding in fronto parietal was explored by the research group led by Prof.Chen Antao(陈安涛)from the Faculty of Psychology,Southwest University with the collaboration of Prof.Tobias Egner from Duke University.The study was supported by the National Natural Science Foundation of China,and was recently published in Journal of Neuroscience(2017,0935-17).
基金Supported by the National Key Research and Development Program of China(2018YFC1507700)National Natural Science Foundation of China(41905083)the United States National Science Foundation(AGS-1419526)。
文摘Two global experiments were carried out to investigate the effects of dynamic vegetation processes on numerical climate simulations from 1948 to 2008.The NCEP Global Forecast System(GFS)was coupled with a biophysical model,the Simplified Simple Biosphere Model(SSi B)version 2(GFS/SSi B2),and it was also coupled with a biophysical and dynamic vegetation model,SSi B version 4/Top-down Representation of Interactive Foliage and Flora Including Dynamics(TRIFFID)(GFS/SSi B4/TRIFFID).The effects of dynamic vegetation processes on the simulation of precipitation,near-surface temperature,and the surface energy budget were identified on monthly and annual scales by assessing the GFS/SSi B4/TRIFFID and GFS/SSi B2 results against the satellite-derived leaf area index(LAI)and albedo and the observed land surface temperature and precipitation.The results show that compared with the GFS/SSiB2 model,the temporal correlation coefficients between the globally averaged monthly simulated LAI and the Global Inventory Monitoring and Modeling System(GIMMS)/Global Land Surface Satellite(GLASS)LAI in the GFS/SSi B4/TRIFFID simulation increased from 0.31/0.29(SSiB2)to 0.47/0.46(SSiB4).The correlation coefficients between the simulated and observed monthly mean near-surface air temperature increased from 0.50(Africa),0.35(Southeast Asia),and 0.39(South America)to 0.56,0.41,and 0.44,respectively.The correlation coefficients between the simulated and observed monthly mean precipitation increased from 0.19(Africa),0.22(South Asia),and 0.22(East Asia)to 0.25,0.27,and 0.28,respectively.The greatest improvement occurred over arid and semiarid areas.The spatiotemporal variability and changes in vegetation and ground surface albedo modeled by the GFS with a dynamic vegetation model were more consistent with the observations.The dynamic vegetation processes contributed to the surface energy and water balance and in turn,improved the annual variations in the simulated regional temperature and precipitation.The dynamic vegetation processes had the greatest influence on the spatiotemporal changes in the latent heat flux.This study shows that dynamic vegetation processes in earth system models significantly improve simulations of the climate mean status.
