基于最优化分块的大规模数据体绘制加速方法

GPU加速体绘制已成为体可视化领域的研究热点，然而超出显存的大规模数据无法直接载入，成为GPU应用的瓶颈。分块技术能够在保证图像质量的条件下解决该问题，但分块数据的频繁加载和访问明显降低了绘制速度。针对上述问题，通过建立最优化分块模型得到了大规模数据的最优分块，并通过构造节点编号纹理和改进距离模板设计的方法进一步提高了基于八叉树的分块体绘制算法的绘制速度。实验结果表明，该方法加速效果明显。

Volume rendering acceleration method based on optimal bricking for large volume data

GPU-based volume rendering has become an active research area in the domain of volume visualization. Large volume data cannot be uploaded directly due to the limitation of GPU memory, which has been a bottleneck of the application of GPU. Bricking method could not only solve this problem, but also maintain the quality of original volume-rendered image. However, the data exchange via the graphics bus is really time consuming and will definitely degrade the render performance. As for these difficulties, the optimal bricking for large volume data was calculated by establishing the model for optimal bricking, and also a 3D texture named node code texture was constructed and distance template was improved to accelerate the octree-based bricking volume rendering. The experimental results illustrate that the proposed method can significantly accelerate the bricking-based volume rendering for large volume data.