提出一种具有较强鲁棒性的残缺点云数据修复算法,借助kD tree寻找点云的缺陷边界,确定点云的缺陷区域;然后利用二次曲面的特性参数化边界点列;最后,通过径向基函数表示的插值曲面计算位于残缺区域内部的数据点,实现残缺点云数据的修复.
k dimension tree is constructed to trace the boundary points of each hole in point cloud, which are refined and parameterized with a normalized quadric surface. RBF surface is defined to interpolate the boundary points and generate uniformly distributed point mesh in the hole, RBF surface is somehow superior to B-spline surface in its shape-preversing and smooth blending ability.