Concepts for a virtual 3D space and a hyper-sphere are proposed and the formulae for determining the computable nodes of the mesh are derived.Then a new optimization design method('Virtual Mesh Method'or V.M.M...Concepts for a virtual 3D space and a hyper-sphere are proposed and the formulae for determining the computable nodes of the mesh are derived.Then a new optimization design method('Virtual Mesh Method'or V.M.M)is developed.Three examples are given,showing that the method proposed is especially suitable for the optimized design of complex structures,and that the global approximate optimal solution can be searched with remarkably reduced computational work.展开更多
Clustering or connected dominating set (CDS) both approaches can establish a virtual backbone (VB) in wireless sensor networks (WSNs) or wireless mesh networks (WMNs). Each cluster consisting of a cluster head (CH) an...Clustering or connected dominating set (CDS) both approaches can establish a virtual backbone (VB) in wireless sensor networks (WSNs) or wireless mesh networks (WMNs). Each cluster consisting of a cluster head (CH) and its neighboring nodes can form a dominating set. After some bridging nodes were selected, cluster heads (CHs) connected through these bridging nodes naturally formed a CDS. Although CDS provides obvious backbone architecture, however, the number of cluster heads and bridging nodes may be too large, this may cause the loss of advantages of virtual backbone. When we effectively reduce their numbers, more effectively WCDS (Weakly Connected Dominating Set) can be fining out. Some essential topics on constructing WCDS-based VB in WSN/WMN are discussed in this paper. From the point of view of three different protocol layers, including network (NWK) layer, MAC layer, and physical (PHY) layer, we explore their cross-layer research topics and design algorithms. For NWK layer, area-based WCDS algorithms and routing strategies including via VB and not via VB are discussed. For MAC layer, a WCDS-based energy-efficient MAC protocol is presented. For PHY layer, battery-aware alternative VB selections and sensor nodes with different transmission ranges are addressed.展开更多
当视频中人物的某些身体部位超出镜头(EFoV,Exceeding Field of View)时,给基于视频的三维人体网格建模任务带来了新的挑战。本文中我们提出视频内细粒度人体部位信息注意力分配(BAA,Body part Attention Allocation)模型框架,通过选用...当视频中人物的某些身体部位超出镜头(EFoV,Exceeding Field of View)时,给基于视频的三维人体网格建模任务带来了新的挑战。本文中我们提出视频内细粒度人体部位信息注意力分配(BAA,Body part Attention Allocation)模型框架,通过选用人体虚拟标记(Virtual Markers)作为三维人体的中间表示,将基于EFoV视频的三维人体网格建模问题转化为人体虚拟标记点集合序列的恢复问题。具体的,首先,我们设计了Rand Mask策略,通过对完整的人体虚拟标记点集合按人体部位进行随机遮蔽,模拟视频中人体部位超出镜头时导致的信息缺失。其次,设计置信度增强过滤模块(CEF,Confidence Enhancedand Filtering),对前置方法估计的虚拟标记置信度进行过滤增强,进而摆脱对前置方法的过度依赖。最后,设计虚拟标记注意力分配模块(AIVM,Attention In Virtual Markers),根据增强后的虚拟标记置信度情况,为各虚拟标记点分配合适的注意力,进而合理地利用序列中虚拟标记的互补关系恢复出完整的虚拟标记序列,并以此估计最终的3D人体网格模型。我们将三维运动捕捉(MoCap)数据集中的SMPL人体网格进行采样,得到对应的虚拟标记点集合数据,人体SMPL模型与虚拟标记点集合共同构成了模型训练中的数据集。展开更多
We present a new algorithm to compute a geodesic path over a triangle mesh. Based on Novotni＇s propagating wavefront method which is similar to the well known Dijkstra algorithm, we made some improvements which Novot...We present a new algorithm to compute a geodesic path over a triangle mesh. Based on Novotni＇s propagating wavefront method which is similar to the well known Dijkstra algorithm, we made some improvements which Novotni had missed and we also gave the method to find out the geodesic path which Novotni had not. It can handle both convex and non-convex surfaces or even with boundaries. Experiment results show that our method works very well both in efficiency and precision.展开更多
Emulating massively parallel computer architectures represents a very important tool for the parallel programmers. It allows them to implement and validate their algorithms. Due to the high cost of the massively paral...Emulating massively parallel computer architectures represents a very important tool for the parallel programmers. It allows them to implement and validate their algorithms. Due to the high cost of the massively parallel real machines, they remain unavailable and not popular in the parallel computing community. The goal of this paper is to present an elaborated emulator of a 2-D massively parallel re-configurable mesh computer of size n x n processing elements (PE). Basing on the object modeling method, we develop a hard kernel of a parallel virtual machine in which we translate all the physical properties of its different components. A parallel programming language and its compiler are also devel-oped to edit, compile and run programs. The developed emulator is a multi platform system. It can be installed in any sequential computer whatever may be its operating system and its processing unit technology (CPU). The size n x n of this virtual re-configurable mesh is not limited;it depends just on the performance of the sequential machine supporting the emulator.展开更多
NOC（network-on-chip）设计中,最重要的问题是如何提高NOC的性能并减小延时。通讯网络中的的节点结构对NOC的性能和延时有着重要影响。而其中通讯节点虚拟通道的buffer深度尤为关键。通过NIRGAM（NOC Interconnect Routing and Ap plic...NOC（network-on-chip）设计中,最重要的问题是如何提高NOC的性能并减小延时。通讯网络中的的节点结构对NOC的性能和延时有着重要影响。而其中通讯节点虚拟通道的buffer深度尤为关键。通过NIRGAM（NOC Interconnect Routing and Ap plication Modeling）仿真器对一个基于XY路由算法的3×4的2D-Mesh结构NOC进行研究。分析结果表明：通讯节点虚拟通道的输入FIFO（First-In-Fist-Out）的buffer深度大于等于6时,NOC即得到优化。而该buffer深度为6到16时,优化效果并不理想。展开更多
基金Project supported by the Natural Science Foundation of Henan Province,China(No.0311010400).
文摘Concepts for a virtual 3D space and a hyper-sphere are proposed and the formulae for determining the computable nodes of the mesh are derived.Then a new optimization design method('Virtual Mesh Method'or V.M.M)is developed.Three examples are given,showing that the method proposed is especially suitable for the optimized design of complex structures,and that the global approximate optimal solution can be searched with remarkably reduced computational work.
文摘Clustering or connected dominating set (CDS) both approaches can establish a virtual backbone (VB) in wireless sensor networks (WSNs) or wireless mesh networks (WMNs). Each cluster consisting of a cluster head (CH) and its neighboring nodes can form a dominating set. After some bridging nodes were selected, cluster heads (CHs) connected through these bridging nodes naturally formed a CDS. Although CDS provides obvious backbone architecture, however, the number of cluster heads and bridging nodes may be too large, this may cause the loss of advantages of virtual backbone. When we effectively reduce their numbers, more effectively WCDS (Weakly Connected Dominating Set) can be fining out. Some essential topics on constructing WCDS-based VB in WSN/WMN are discussed in this paper. From the point of view of three different protocol layers, including network (NWK) layer, MAC layer, and physical (PHY) layer, we explore their cross-layer research topics and design algorithms. For NWK layer, area-based WCDS algorithms and routing strategies including via VB and not via VB are discussed. For MAC layer, a WCDS-based energy-efficient MAC protocol is presented. For PHY layer, battery-aware alternative VB selections and sensor nodes with different transmission ranges are addressed.
文摘当视频中人物的某些身体部位超出镜头(EFoV,Exceeding Field of View)时,给基于视频的三维人体网格建模任务带来了新的挑战。本文中我们提出视频内细粒度人体部位信息注意力分配(BAA,Body part Attention Allocation)模型框架,通过选用人体虚拟标记(Virtual Markers)作为三维人体的中间表示,将基于EFoV视频的三维人体网格建模问题转化为人体虚拟标记点集合序列的恢复问题。具体的,首先,我们设计了Rand Mask策略,通过对完整的人体虚拟标记点集合按人体部位进行随机遮蔽,模拟视频中人体部位超出镜头时导致的信息缺失。其次,设计置信度增强过滤模块(CEF,Confidence Enhancedand Filtering),对前置方法估计的虚拟标记置信度进行过滤增强,进而摆脱对前置方法的过度依赖。最后,设计虚拟标记注意力分配模块(AIVM,Attention In Virtual Markers),根据增强后的虚拟标记置信度情况,为各虚拟标记点分配合适的注意力,进而合理地利用序列中虚拟标记的互补关系恢复出完整的虚拟标记序列,并以此估计最终的3D人体网格模型。我们将三维运动捕捉(MoCap)数据集中的SMPL人体网格进行采样,得到对应的虚拟标记点集合数据,人体SMPL模型与虚拟标记点集合共同构成了模型训练中的数据集。
基金This work was supported by National Natural Science Foundation of PRC(No.60503058,No.60533080)the Science and Technology Fund of Huawei Technologies Co.,Ltd.
文摘We present a new algorithm to compute a geodesic path over a triangle mesh. Based on Novotni＇s propagating wavefront method which is similar to the well known Dijkstra algorithm, we made some improvements which Novotni had missed and we also gave the method to find out the geodesic path which Novotni had not. It can handle both convex and non-convex surfaces or even with boundaries. Experiment results show that our method works very well both in efficiency and precision.
文摘Emulating massively parallel computer architectures represents a very important tool for the parallel programmers. It allows them to implement and validate their algorithms. Due to the high cost of the massively parallel real machines, they remain unavailable and not popular in the parallel computing community. The goal of this paper is to present an elaborated emulator of a 2-D massively parallel re-configurable mesh computer of size n x n processing elements (PE). Basing on the object modeling method, we develop a hard kernel of a parallel virtual machine in which we translate all the physical properties of its different components. A parallel programming language and its compiler are also devel-oped to edit, compile and run programs. The developed emulator is a multi platform system. It can be installed in any sequential computer whatever may be its operating system and its processing unit technology (CPU). The size n x n of this virtual re-configurable mesh is not limited;it depends just on the performance of the sequential machine supporting the emulator.
文摘NOC（network-on-chip）设计中,最重要的问题是如何提高NOC的性能并减小延时。通讯网络中的的节点结构对NOC的性能和延时有着重要影响。而其中通讯节点虚拟通道的buffer深度尤为关键。通过NIRGAM（NOC Interconnect Routing and Ap plication Modeling）仿真器对一个基于XY路由算法的3×4的2D-Mesh结构NOC进行研究。分析结果表明：通讯节点虚拟通道的输入FIFO（First-In-Fist-Out）的buffer深度大于等于6时,NOC即得到优化。而该buffer深度为6到16时,优化效果并不理想。
