摄像机与激光测距仪(Camera and laser rangefinder,LRF)被广泛应用于机器人、移动道路测量车、无人驾驶等领域.其中,外参数标定是实现图像与LIDAR数据融合的第一步,也是至关重要的一步.本文提出一种新的基于最小解(Minimal solution)...摄像机与激光测距仪(Camera and laser rangefinder,LRF)被广泛应用于机器人、移动道路测量车、无人驾驶等领域.其中,外参数标定是实现图像与LIDAR数据融合的第一步,也是至关重要的一步.本文提出一种新的基于最小解(Minimal solution)外参数标定算法,即摄像机与激光仅需对标定棋盘格采集三次数据.本文首次提出虚拟三面体概念,并以之构造透视三点问题(Perspective-three-point,P3P)用以计算激光与摄像机之间的坐标转换关系.相对于文献在对偶三维空间(Dual 3D space)中构造的P3P问题,本文直接在原始三维空间中构造P3P问题,具有更直观的几何意义,更利于对P3P问题进行求解与分析.针对P3P问题多达八组解的问题,本文还首次提出一种平面物成像区域约束方法从多解中获取真解,使得最小解标定法具有更大的实用性与灵活性.实验中分别利用模拟数据与真实数据对算法进行测试.算法结果表明,在同等输入的条件下,本文算法性能超过文献中的算法.本文所提的平面物成像区域约束方法能从多解中计算出真解,大大提高了最小解算法的实用性与灵活性.展开更多
In the PnP problem,the imaging devices follow the perspective rule and the imaging rays pass through a common point. However,there are many new imaging devices being developed for robot navigation or other fields with...In the PnP problem,the imaging devices follow the perspective rule and the imaging rays pass through a common point. However,there are many new imaging devices being developed for robot navigation or other fields with the advance in imaging technologies for machine vision. These devise are not necessarily being designed to follow the perspective rule in order to satisfy some design criterion and,thus, the imaging rays may not pass through a common point.Such generalized imaging devices may not be perspective and, therefore, their poses cannot be estimated with traditional perspective technique.Using the Wu-Ritt＇s zero decomposition method,the main component for the nonperspective-three-point problem is given. We prove that there are at most eight solutions in the general case and give the solution classification for the NP3P problem.展开更多
文摘摄像机与激光测距仪(Camera and laser rangefinder,LRF)被广泛应用于机器人、移动道路测量车、无人驾驶等领域.其中,外参数标定是实现图像与LIDAR数据融合的第一步,也是至关重要的一步.本文提出一种新的基于最小解(Minimal solution)外参数标定算法,即摄像机与激光仅需对标定棋盘格采集三次数据.本文首次提出虚拟三面体概念,并以之构造透视三点问题(Perspective-three-point,P3P)用以计算激光与摄像机之间的坐标转换关系.相对于文献在对偶三维空间(Dual 3D space)中构造的P3P问题,本文直接在原始三维空间中构造P3P问题,具有更直观的几何意义,更利于对P3P问题进行求解与分析.针对P3P问题多达八组解的问题,本文还首次提出一种平面物成像区域约束方法从多解中获取真解,使得最小解标定法具有更大的实用性与灵活性.实验中分别利用模拟数据与真实数据对算法进行测试.算法结果表明,在同等输入的条件下,本文算法性能超过文献中的算法.本文所提的平面物成像区域约束方法能从多解中计算出真解,大大提高了最小解算法的实用性与灵活性.
基金This project was partially supported by Shuxue Tianyuan Foundation(No.10526031).
文摘In the PnP problem,the imaging devices follow the perspective rule and the imaging rays pass through a common point. However,there are many new imaging devices being developed for robot navigation or other fields with the advance in imaging technologies for machine vision. These devise are not necessarily being designed to follow the perspective rule in order to satisfy some design criterion and,thus, the imaging rays may not pass through a common point.Such generalized imaging devices may not be perspective and, therefore, their poses cannot be estimated with traditional perspective technique.Using the Wu-Ritt＇s zero decomposition method,the main component for the nonperspective-three-point problem is given. We prove that there are at most eight solutions in the general case and give the solution classification for the NP3P problem.
