摘要
本文介绍了一种利用移动机器人的激光信息和立体视觉信息实现室内外场景三维重建的方法。推导了激光与视觉采样点的配准关系。通过改变黑色三角形标定板的位置和姿态,选取足够的观测次数以获得足够的观测样本,使用高斯牛顿迭代方法进行估计,得到二维激光扫描仪与立体相机之间的变换矩阵。使用双目立体视觉里程计进行定位,采用八叉树算法存储和查找点,删除冗余点,得到场景的三维模型显示。实验证明本文所提出的方法有效地解决了室内外场景三维重建的基本问题。
This paper presents a method of creating 3D models of indoor and outdoor scenes from laser and stereo visual information of a mobile robot. We deduce the formulas of registration between laser points and visual points. By changing the position and gesture of the black triangle calibration plate, we select a sufficient number of observations to obtain a sufficient sample of observations, using the Gauss-Newton iterative method to estimate the transformation matrix between the two-dimensional laser scanner and stereo camera. We use binocular stereo vision odometer positioning. Using octree algorithm to store, find points and remove redundant points, we get the three-dimensional model of the scene shows. The experiments show that the proposed method can effectively solve the basic problem of the three-dimensional reconstruction of the indoor and outdoor scenes.
出处
《软件》
2012年第11期1-5,共5页
Software
基金
国家自然科学基金资助项目(50775013)
高等学校科技创新工程重大项目培育资金资助(708011)
关键词
激光与视觉同步数据
联合标定
纹理映射
三维重建
synchronized laser and visual data
joint calibration
texture mapping
3D modeling
