欢迎访问《图学学报》 分享到:

图学学报

• 图形学与可视化 • 上一篇    下一篇

基于最近离散点的光线跟踪

  

  • 出版日期:2013-06-29 发布日期:2015-06-11

A Ray Tracing Method Based on the Nearest Points

  • Online:2013-06-29 Published:2015-06-11

摘要: :论文提出基于最近离散点的光线跟踪算法。对于原始的点云模型,算法通
过平衡二叉树在设定范围内搜索离光线迭代点最近的N 个离散点,并用栅格的加速结构避
免不必要的迭代搜索计算。光线与实际最近的N′个离散点的局部平面进行求交计算,局部
平面的法向量是由离散点对应的三角形法向量通过三角形面积加权平均计算得到,并保证法
向量计算的一致性。通过改变光线跟踪的参数(最近离散点的数目),即可达到渐进地多分
辨率显示原始的点云模型的目的。对于噪声多的原始的点云模型,设置较大的最近离散点的
数目,以有效地减少其绘制的噪声;对于噪声少的原始的点云模型,设置较小的最近离散点
的数目,以更多地显示其局部几何特征。

关键词: 点云, 光线跟踪, 最近的N 个离散点, 平衡二叉树

Abstract: This paper presents a ray tracing method based on the nearest points. Based on
rude point clouds, we employ a balanced binary tree to locate the nearest N points in setting range,
and employ grid structure to avoid unnecessarily iterative computation in point clouds space. The
intersecting computation is done between ray and local plane. To the actual nearest N’ points, the
normal of local plane is computed by triangle-area-weighted average computation of normals of
corresponding triangles, and we ensure coherence of computation of normal. Our method can
progressively render rude point clouds in multi-resolution mode by changing ray tracing
parameter (number of nearest points), and we set bigger number of nearest points to effectively
remove noise of rendering for point clouds with many noises, and we set smaller number of
nearest points to show more local geometry character for point clouds with less noise.

Key words: point clouds, ray tracing, the nearest N points, balanced binary tree