Mobile version of Hanspub

文章引用说明 更多>> (返回到该文章)

Deng, Y. and Lin, X. (2006) A fast line segment based dense stereo algorithm using tree dynamic pro-gramming. Proceedings of 9th European Conference on Computer Vision, Graz, 7-13 May 2006, 201-212.


  • 标题: 采用彩色分割立体匹配与简化点云的三维目标快速重建Fast Reconstruction of 3-D Object Based on Color Image Segmentation Stereo Matching and Point Cloud Reduction

    作者: 李鹤喜, 张娟娟, 孙玲云

    关键字: 均值漂移, 图像分割, 立体匹配, 点云, 三维重建Mean Shift, Image Segmentation, Stereo Matching, Point Cloud, 3D Reconstruction

    期刊名称: 《Artificial Intelligence and Robotics Research》, Vol.3 No.4, 2014-11-28

    摘要: 本文将彩色分割立体匹配和简化点云技术用于三维目标的快速重建。对于采集的左右两幅三维目标图像,首先采用均值漂移算法进行彩色图像分割,然后按区域匹配算法进行快速立体匹配,获得初始视差,再应用置信传播法进行全局视差优化,从而得到精确的视差图与空间点云。应用空间表面曲率准则对获取的密集点云进行简化,并采用Delaunay三角剖分算法进行三维重建。实验结果表明:采用彩色图像分割与置信传播相结合,能够改善立体匹配效率并保证了匹配质量;基于曲率的点云简化,既提高了三维重建速度也得到了满意的重建效果。Color image segmentation stereo matching and point cloud reduction method is used for fast re-construction of 3-dimensional (3-D) object in this paper. For two captured images of a 3-D object, color image segmentation is first carried out using mean shift algorithm and initial disparity is computed using fast region-based stereo matching, and then the accurate disparity and point cloud of the 3-D object are obtained using belief propagation method to optimize global disparity. The 3-D object is reconstructed using Delaunay triangulation algorithm and point cloud reduction processing based on a surface curvature criterion. The experimental results show that the combi-nation of color image segmentation with belief propagation method can improve stereo matching efficiency and ensure matching quality, and the point cloud reduction technique can rise 3D re-construction speed and obtain satisfactory 3-D reconstruction result.