一种基于八叉树的快速体素化方法

A fast voxelization method based on octree

当前的体素化方法大都具有较高的复杂性,计算开销大,对硬件要求高。为了简单快速地实现3D模型的体素化,提出一种基于八叉树的快速体素化方法。首先在使用八叉树进行模型细分的基础上,得到模型表面数据。然后根据表面数据选择多个方向对模型进行逐行扫描。该方法能快速地区分出模型的外部数据和内部数据,最终实现模型的体素化。对不同分辨率下的多种模型的实验结果表明,文中提出的方法能有效地实现快速体素化,具有一定的应用价值。

Current methods of voxelization almost have extraordinary complexity which need a high cost of calculation and hardware requirements. In order to achieve a simple and fast voxelization of 3D models,this paper presents a fast voxelization method based on octree. Firstly,subdividing the 3D models using octree structure to get the voxel data on the model surface. After selecting multi-directions to scan the model per row according to the surface data,it's easy to judge that the voxels are inside of the model or outside of the model. Then the voxelization of 3D model is done. The experimental results of several models on different resolutions show that this method achieves a fast voxelization efficiently and is valuable in application.