Using elastic crystalline viscoplastic finite element (FE) annlysis, the formability of BCC steel sheets was assessed. An orientation probability assignment method in the FE modeling procedure, which can be categorize...Using elastic crystalline viscoplastic finite element (FE) annlysis, the formability of BCC steel sheets was assessed. An orientation probability assignment method in the FE modeling procedure, which can be categorized as an inhomogenized material modeling, was newly proposed. In the study, the crystal orientations of three materials, mild steel, dual phase steel and the high strength steel, were obtained by Xray diffraction and orientation distribution function (ODF) analyses. The measured ODF results have revealed clearly different textures in the sheets, featured by orientation fibers, skeleton lines and selected orientations in Euler angle space, which are closely related to the plastic anisotropy. Then, the crystal orientations were assigned to FE integration points by using this ODF data, individually. The FE analyses of the standard limiting dome height (LDH) test show how the fiber textures affect the extent of strain localization in the forming processes. It was confirmed by comparison with experimental results that this FE code could predict the extreme strain localization and assess the sheet formability.展开更多
在视觉同时定位与地图构建问题中,ORB(Oriented FAST and Rotated BRIEF)特征由于其高效、稳定的优点而受到广泛关注。针对ORB特征提取过程中存在的像点量测精度较低、特征聚集现象明显等问题,提出了一种适用于高精度SLAM的均衡化亚像素...在视觉同时定位与地图构建问题中,ORB(Oriented FAST and Rotated BRIEF)特征由于其高效、稳定的优点而受到广泛关注。针对ORB特征提取过程中存在的像点量测精度较低、特征聚集现象明显等问题,提出了一种适用于高精度SLAM的均衡化亚像素ORB特征提取方法。分析了精确特征定位的原理,对误差方程进行合理的简化并采用一种基于模板窗口距离的权函数计算方法,大幅降低了计算负担;设计了一种基于四叉树结构的特征均衡化方案,对包含特征的像平面空间进行有限次数的迭代分割,然后选取具有最优响应的特征。试验表明,本文方法进行特征提取的额外计算负担小于2.5 ms,在运行TUM和KITTI数据集时,ORB特征的量测精度分别为0.84和0.62 Pixel,达到亚像素水平,可以降低误差初值,提高光束法平差效率,并能够在满足特征总体分布规律的情况下,显著改善特征聚集的现象,有利于后续问题的稳健、准确求解。展开更多
基于相机的无人驾驶汽车视觉同步定位与地图构建(SLAM),可完成无人驾驶汽车的定位与建图。针对传统ORB(Oriented FAST and Rotated BRIEF)算法在提取图像特征点时容易造成冗杂、分布集中的问题,提出一种限制四叉树算法分裂深度的改进ORB...基于相机的无人驾驶汽车视觉同步定位与地图构建(SLAM),可完成无人驾驶汽车的定位与建图。针对传统ORB(Oriented FAST and Rotated BRIEF)算法在提取图像特征点时容易造成冗杂、分布集中的问题,提出一种限制四叉树算法分裂深度的改进ORB(A-ORB)算法。该算法构造图像金字塔解决尺度不变性问题;根据所提取的特征点总数计算出每层金字塔所需要提取的特征点数;对每层金字塔图像采用自适应区域划分,根据图像信息计算特征点提取阈值;利用改进四叉树算法来均匀化分布特征点。进行了模拟实验。结果表明:相较于ORB、MA以及S-ORB算法,该算法运行效率提高了30%以上,匹配精度提高了10%以上。展开更多
基金the National Natural Science Foundation of China for financial support(Grant No.59875025) to the research cooperation with OIT,Japan
文摘Using elastic crystalline viscoplastic finite element (FE) annlysis, the formability of BCC steel sheets was assessed. An orientation probability assignment method in the FE modeling procedure, which can be categorized as an inhomogenized material modeling, was newly proposed. In the study, the crystal orientations of three materials, mild steel, dual phase steel and the high strength steel, were obtained by Xray diffraction and orientation distribution function (ODF) analyses. The measured ODF results have revealed clearly different textures in the sheets, featured by orientation fibers, skeleton lines and selected orientations in Euler angle space, which are closely related to the plastic anisotropy. Then, the crystal orientations were assigned to FE integration points by using this ODF data, individually. The FE analyses of the standard limiting dome height (LDH) test show how the fiber textures affect the extent of strain localization in the forming processes. It was confirmed by comparison with experimental results that this FE code could predict the extreme strain localization and assess the sheet formability.
基金Supposed by the National Natural Science Foundation of China under Grant No.60575019(国家自然科学基金)the National High-Tech Research and Development Plan of China under Grant No.2006AA01Z116(国家高技术研究发展计划(863))
文摘在视觉同时定位与地图构建问题中,ORB(Oriented FAST and Rotated BRIEF)特征由于其高效、稳定的优点而受到广泛关注。针对ORB特征提取过程中存在的像点量测精度较低、特征聚集现象明显等问题,提出了一种适用于高精度SLAM的均衡化亚像素ORB特征提取方法。分析了精确特征定位的原理,对误差方程进行合理的简化并采用一种基于模板窗口距离的权函数计算方法,大幅降低了计算负担;设计了一种基于四叉树结构的特征均衡化方案,对包含特征的像平面空间进行有限次数的迭代分割,然后选取具有最优响应的特征。试验表明,本文方法进行特征提取的额外计算负担小于2.5 ms,在运行TUM和KITTI数据集时,ORB特征的量测精度分别为0.84和0.62 Pixel,达到亚像素水平,可以降低误差初值,提高光束法平差效率,并能够在满足特征总体分布规律的情况下,显著改善特征聚集的现象,有利于后续问题的稳健、准确求解。
文摘基于相机的无人驾驶汽车视觉同步定位与地图构建(SLAM),可完成无人驾驶汽车的定位与建图。针对传统ORB(Oriented FAST and Rotated BRIEF)算法在提取图像特征点时容易造成冗杂、分布集中的问题,提出一种限制四叉树算法分裂深度的改进ORB(A-ORB)算法。该算法构造图像金字塔解决尺度不变性问题;根据所提取的特征点总数计算出每层金字塔所需要提取的特征点数;对每层金字塔图像采用自适应区域划分,根据图像信息计算特征点提取阈值;利用改进四叉树算法来均匀化分布特征点。进行了模拟实验。结果表明:相较于ORB、MA以及S-ORB算法,该算法运行效率提高了30%以上,匹配精度提高了10%以上。
