针对铁路异物侵限频繁发生导致的列车运行安全问题,提出一种基于背景感知相关滤波器的铁路异物侵限跟踪方法。利用方向梯度直方图(HOG,Histogram of Oriented Gradient)特征提取铁路侵限异物自身特征,结合剪裁矩阵,以增加视频帧中实际...针对铁路异物侵限频繁发生导致的列车运行安全问题,提出一种基于背景感知相关滤波器的铁路异物侵限跟踪方法。利用方向梯度直方图(HOG,Histogram of Oriented Gradient)特征提取铁路侵限异物自身特征,结合剪裁矩阵,以增加视频帧中实际背景的负样本;使用交替方向乘子法(ADMM,Alternating Direction Method of Multipliers)训练背景感知相关滤波器,减少计算复杂度,在保证跟踪速度的前提下,提升跟踪侵限异物的准确性,从而适应铁路沿线环境中由于侵限异物的形变、快速移动或天气等原因造成的目标丢失及跟踪框漂移等情况。实验结果表明,该方法对铁路侵限异物的跟踪精确度和AUC(Area Under Curve)值分别达到93%和71.9%,均高于SRDCF、KCF、ASLA和CSK等算法,具有更好的准确性。展开更多
The current study performed a finite element analysis of the strain localization behavior of a voided ductile material using a non-local plasticity formulation in which the yield strength depends on both an equivalent...The current study performed a finite element analysis of the strain localization behavior of a voided ductile material using a non-local plasticity formulation in which the yield strength depends on both an equivalent plastic strain measurement (hardening parameter) and Laplacian equivalent. The introduction of gradient terms to the yield function was found to play an important role in simulating the strain localization behavior of the voided ductile material. The effect of the mesh size and characteristic length on the strain localization were also investigated. An FEM simulation based on the proposed non-local plasticity revealed that the load-strain curves of the voided ductile material subjected to plane strain tension converged to one curve, regardless of the mesh size. In addition, the results using non-local plasticity also exhibited that the dependence of the deformation behavior of the material on the mesh size was much less sensitive than that with classical local plasticity and could be successfully eliminated through the introduction of a large value for the characteristic length.展开更多
文摘针对铁路异物侵限频繁发生导致的列车运行安全问题,提出一种基于背景感知相关滤波器的铁路异物侵限跟踪方法。利用方向梯度直方图(HOG,Histogram of Oriented Gradient)特征提取铁路侵限异物自身特征,结合剪裁矩阵,以增加视频帧中实际背景的负样本;使用交替方向乘子法(ADMM,Alternating Direction Method of Multipliers)训练背景感知相关滤波器,减少计算复杂度,在保证跟踪速度的前提下,提升跟踪侵限异物的准确性,从而适应铁路沿线环境中由于侵限异物的形变、快速移动或天气等原因造成的目标丢失及跟踪框漂移等情况。实验结果表明,该方法对铁路侵限异物的跟踪精确度和AUC(Area Under Curve)值分别达到93%和71.9%,均高于SRDCF、KCF、ASLA和CSK等算法,具有更好的准确性。
文摘The current study performed a finite element analysis of the strain localization behavior of a voided ductile material using a non-local plasticity formulation in which the yield strength depends on both an equivalent plastic strain measurement (hardening parameter) and Laplacian equivalent. The introduction of gradient terms to the yield function was found to play an important role in simulating the strain localization behavior of the voided ductile material. The effect of the mesh size and characteristic length on the strain localization were also investigated. An FEM simulation based on the proposed non-local plasticity revealed that the load-strain curves of the voided ductile material subjected to plane strain tension converged to one curve, regardless of the mesh size. In addition, the results using non-local plasticity also exhibited that the dependence of the deformation behavior of the material on the mesh size was much less sensitive than that with classical local plasticity and could be successfully eliminated through the introduction of a large value for the characteristic length.
