• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International symposium on photoelectronic detection and imaging;ISPDI 2011 >A Quaternion Pose Determination Solution Based on Monocular Vision Model
【24h】

A Quaternion Pose Determination Solution Based on Monocular Vision Model

机译：基于单目视觉模型的四元数姿势确定解决方案

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Determination of relative three-dimensional position and orientation between two reference frames can be solved by the pose measuring methods based on monocular vision model. Owing to the special T-shaped configuration, the definition of object rotational matrix in the terms of quaternion elements helped in representing the problem by six nonlinear equations from which a closed-form solution can be obtained for all the unknown parameters. The calculating formulas of elements in the rotational matrix were deduced from the coordinates of feature points in camera frame as well as the converting vector which was also introduced into the process acting as corrected term. An approximate pose could be found by the assumption of zero difference in depth of all points in camera frame, then the converting vector should be initialized by the third row of current rotational matrix. The principle of computing priority of the max value in quaternion expression was proposed to ensure the convergence of the iteration loop through which the final pose was achieved in a few iterations. Simulation experiments show the validity of the solution and analysis of the calculating precision was made in detail. The measuring orientation error would constringe with the reduction of distance from camera focus to target object and performance of the algorithm went well in short distance, while the deformation went larger with the increasing of errors caused by imprecise correspondence.
机译：可以通过基于单眼视觉模型的姿态测量方法来解决两个参考系之间的相对三维位置和方向的确定。由于具有特殊的T形构造，用四元数元素定义对象旋转矩阵有助于通过六个非线性方程式来表示问题，从中可以得到所有未知参数的闭式解。从摄像机帧中特征点的坐标以及转换矢量推导出旋转矩阵中各元素的计算公式，并将其作为校正项引入到处理中。可以通过假设摄像机帧中所有点的深度差为零来找到近似姿势，然后应通过当前旋转矩阵的第三行初始化转换矢量。提出了计算四元数表达式中最大值的优先级的原理，以确保迭代循环的收敛性，从而在几次迭代中获得最终姿势。仿真实验表明了该方法的有效性，并对计算精度进行了详细分析。测量方向误差会随着相机焦点到目标物体距离的减小而收敛，并且算法的性能在短距离内表现良好，而变形则随着不精确对应引起的误差的增加而增大。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号