基于自适应八叉树划分的高精度四面体可视化

摘要

Converting a tetrahedral volumetric data into regular volumetric data by an octree can improve the interactivity of the system effectively. When the depth of the octree is higher, the rendering results will be better. However, memory consumption and processing time will also increase. This paper proposes an adaptive regularization reformulation algorighm to construct the octree, improving the original single sampling strategy, combining with the depth information to transfer the sampling results into an octree texture which allows for random access in GPU. Then we use ray casting algorighm to render the regular volumetric data. Because of varying characteristics of the regional depth, the sampling algorithm responds with different step-size strategy. The experimental results show that this method reduces the memory consumption and processing time of the data, and at the same time improves the rendering quality as well as the rendering efficiency.%利用八叉树结构将四面体数据转化为规则网格数据，能有效提高系统的交互性能。八叉树的划分层次越高，绘制效果越好，但数据的存储空间以及处理时间也将大幅增多。提出自适应的规则化表示方法来构建八叉树结构，改进原有的单一采样策略，并结合深度信息将采样结果转换成适用于 GPU 的八叉树纹理结构。然后采用光线投射算法来对体数据进行绘制，根据各区域深度不一的特点，提出了变步长的采样绘制策略。实验结果表明，本文方法降低了数据的空间存储量和处理时间，同时在绘制质量、绘制效率方面都得到了较大提高。