Debris flows pose serious risks to communities in mountainous areas,often resulting in large losses of human life and property.The impeding presence of urban buildings often affects the runout behavior and deposition ...Debris flows pose serious risks to communities in mountainous areas,often resulting in large losses of human life and property.The impeding presence of urban buildings often affects the runout behavior and deposition of debris flows.But the impact of different building densities and sizes on debris flow dynamics has yet to be quantified to guide urban planning in debris flow risk zones.This study focused on a debris flow that occurred in Zhouqu County,Gansu Province,China on August 7th,2010,which was catastrophic and destroyed many buildings.The FLO-2D software was used to simulate this debris flow in two scenarios,i.e.the presence and the absence of buildings,to obtain debris-flow intensity parameters.The developed model was then used to further analyze the influence of large buildings and narrow channels within the urban environment.The simulation results show that considering the presence of buildings in the simulation is essential for accurate assessment of debris flow intensity and deposition distribution.The layout of buildings in the upstream urban area,such as large buildings or parallel buildings which form narrow channels,can affect the flow velocity and depth of debris flow heading towards downstream buildings.To mitigate damage to downstream buildings,the relative spacing(d/a)between upstream and downstream buildings should not exceed a value of two and should ideally be even lower.These findings provide valuable insights for improving the resistance of mountainous cities to urban debris flows.展开更多
移动主体获得准确的定位信息是构建稳定的混合现实(mixed reality,MR)系统的关键,然而MR中的前景对象对传统定位算法的精度影响较大.现阶段基于深度学习的定位算法可以通过识别前景对象来提升精度,但深度学习模型耗时过高,导致算法实时...移动主体获得准确的定位信息是构建稳定的混合现实(mixed reality,MR)系统的关键,然而MR中的前景对象对传统定位算法的精度影响较大.现阶段基于深度学习的定位算法可以通过识别前景对象来提升精度,但深度学习模型耗时过高,导致算法实时性下降.针对该问题,提出了一种MR中融合语义特征传播模型的前景对象感知定位算法.该算法依托语义分割网络与一种快速旋转的二进制独立稳定描述子特征(oriented fast and rotated binary robust independent elementary feature,ORB)提取算法构建了语义特征传播模型,实现高速语义特征提取;融合该模型和几何特征检测方法实现算法中的前景对象感知层,并依赖该感知层剔除MR中前景对象的特征点,构建了背景特征点集,实现高精度、高实时性的定位.实验结果表明:在慕尼黑工业大学(Technical University of Munich,TUM)公共数据集的高动态前景对象场景中,相比动态语义视觉同步定位与建图(dynamic semantic visual simultaneous localization and mapping,DS-SLAM)算法,该算法相对位姿误差降低了60.5%,定位实时性提升了39.5%,可见该算法在MR中具有较高的应用价值.展开更多
基金This study was funded by the National Key Research and Development Program of China(Grant No.2019YFC1806001)the National Natural Science Foundation of China(Grant No.51988101,Grant No.52278376,Grant No.42007245)the Science and Technology Development Fund,Macao SAR(File nos.0083/2020/A2 and 001/2024/SKL).
文摘Debris flows pose serious risks to communities in mountainous areas,often resulting in large losses of human life and property.The impeding presence of urban buildings often affects the runout behavior and deposition of debris flows.But the impact of different building densities and sizes on debris flow dynamics has yet to be quantified to guide urban planning in debris flow risk zones.This study focused on a debris flow that occurred in Zhouqu County,Gansu Province,China on August 7th,2010,which was catastrophic and destroyed many buildings.The FLO-2D software was used to simulate this debris flow in two scenarios,i.e.the presence and the absence of buildings,to obtain debris-flow intensity parameters.The developed model was then used to further analyze the influence of large buildings and narrow channels within the urban environment.The simulation results show that considering the presence of buildings in the simulation is essential for accurate assessment of debris flow intensity and deposition distribution.The layout of buildings in the upstream urban area,such as large buildings or parallel buildings which form narrow channels,can affect the flow velocity and depth of debris flow heading towards downstream buildings.To mitigate damage to downstream buildings,the relative spacing(d/a)between upstream and downstream buildings should not exceed a value of two and should ideally be even lower.These findings provide valuable insights for improving the resistance of mountainous cities to urban debris flows.
文摘移动主体获得准确的定位信息是构建稳定的混合现实(mixed reality,MR)系统的关键,然而MR中的前景对象对传统定位算法的精度影响较大.现阶段基于深度学习的定位算法可以通过识别前景对象来提升精度,但深度学习模型耗时过高,导致算法实时性下降.针对该问题,提出了一种MR中融合语义特征传播模型的前景对象感知定位算法.该算法依托语义分割网络与一种快速旋转的二进制独立稳定描述子特征(oriented fast and rotated binary robust independent elementary feature,ORB)提取算法构建了语义特征传播模型,实现高速语义特征提取;融合该模型和几何特征检测方法实现算法中的前景对象感知层,并依赖该感知层剔除MR中前景对象的特征点,构建了背景特征点集,实现高精度、高实时性的定位.实验结果表明:在慕尼黑工业大学(Technical University of Munich,TUM)公共数据集的高动态前景对象场景中,相比动态语义视觉同步定位与建图(dynamic semantic visual simultaneous localization and mapping,DS-SLAM)算法,该算法相对位姿误差降低了60.5%,定位实时性提升了39.5%,可见该算法在MR中具有较高的应用价值.
