基于双目视觉的三维重建和拼接技术研究

发布时间：2019-06-18 09:19

【摘要】：针对自由移动双目视觉的三维重建问题,提出了一种稠密匹配和稀疏特征匹配相结合的三维重建和点云拼接技术。首先,在对极几何约束的条件下,采用绝对值差和作为匹配度量函数对双目视觉左右图像上的图像点进行逐一匹配,并结合双目视觉内外参数获得各个位置处的稠密三维点云坐标。其次,对相邻两次移动位置处采集的双目视觉图像进行SURF(Speed-up robust features)特征提取和匹配,分别计算匹配SURF特征点在不同位置下的三维坐标,进而求解两个采集位置之间的旋转矩阵和平移矢量,以实现相邻两位置三维点云的坐标统一。最后,在室内采用双目视觉实现了两个视点数据的三维重建和拼接。实验结果表明,文中算法具有计算简单、点云拼接精度较高的优点,可以应用于大场景三维数据的重建工作。
[Abstract]:In order to solve the problem of three-dimensional reconstruction of the free-moving binocular vision, a three-dimensional reconstruction and point cloud splicing technique with a combination of dense matching and sparse feature matching is proposed. First, under the condition of extreme geometric constraint, the image points on the left and right images of the binocular vision are matched one by one by using the absolute value difference and the matching measure function, and the dense three-dimensional point cloud coordinates at various positions are obtained by combining the binocular visual internal and external parameters. secondly, the feature extraction and matching of the SURF feature points are carried out on the binocular vision images acquired at the two adjacent mobile positions, the three-dimensional coordinates of the matched SURF feature points in different positions are calculated respectively, and the rotation matrix and the translation vector between the two acquisition positions are solved, So as to realize the coordinate unification of the three-dimensional point cloud at two adjacent positions. And finally, the three-dimensional reconstruction and the splicing of the two viewpoint data are realized by adopting the binocular vision in the room. The experimental results show that the algorithm has the advantages of simple calculation and high splicing accuracy of point cloud, and can be applied to the reconstruction of three-dimensional data of large scene.
【作者单位】：长春理工大学光电工程学院;
【基金】：吉林省科技发展计划资助青年基金项目(20160520018JH) 吉林省重点科技攻关项目(20120357)
【分类号】：TP391.41

【相似文献】