基于空间活动轮廓模型的缺陷点云重构

工程实际中的点云缺陷使得传统的基于显式曲面的重构方法需要大量的人工操作和修补。文章提出了一种以隐式曲面为基础,无需任何人工操作和点云附加信息的重构方法。将活动轮廓模型扩展到三维,提出了加快迭代和提高收敛质量的自适应新算法,利用改进的水平集数值求解算法获得最终曲面。数值实验说明此方法能以较快的收敛速度重构出高质量复杂曲面。

Surface Reconstruction of Defective Point Clouds Based on Space Active Contour Model

In practice, the defective point clouds often lead to massive artificial work following the explicit surface based reconstruction method. This paper proposes a new method based on implicit surface without any manual work and extra data, which extends the active contour model into 3D space by defining new adaptive convergence and acceleration function. The level set algorithm is improved for numerical implementation, and the numerical experiments demonstrate that the proposed new method can generate high quality surface with faster convergence speed.