高级检索

基于多视图约束的结构光扫描系统冗余点去除算法

An Algorithm for Removing Redundancy Points in Structured Light Scanning System Based on Multi-view Constraints

  • 摘要: 针对结构光多视扫描拼接后在扫描重叠区域内重复计算带来的数据冗余问题,提出一种重复区域内冗余点删除算法.首先根据多视图几何约束原理,将不同视下相机外参数绝对定向.然后计算当前视下三维点并将其反投影至上次扫描图像,若反投影像点在上次扫描中曾参与物点计算,且若2次物点距离大于给定阈值,则保留当前视下三维点,反之保留反投影残差较小者;否则,保留该点.理论分析和实验结果表明,对于重叠扫描区域,该算法能达到和单次扫描相同密度的点云,且去除了扫描过程中精度较低的点,同时保留了曲率较大处表面的细节信息.

     

    Abstract: A novel algorithm has been proposed to remove the redundancy points in the overlapped regions when structured light scanning system works at different angles.All the external parameters in different views are oriented absolutely to a global coordinate system, and the 3D points in current view are reprojected to the last scanned image.If the reprojected image points are triangularized in last scan process, and the distance between the points between current scan and the last scan is larger than agiven threshold value, the 3D points of current scan are preserved.Otherwise, the points with small reprojected residual are preserved.If the reprojected image points are not triangularized in last scan, 3D points in current view are preserved.Both theoretical analysis and experimental results demonstrate that the new algorithm has the benefit of removing the lower accuracy points and preserving more detailed information of the surface with large curvature, with the total data density remaining the same as single view scanning.

     

/

返回文章
返回