摘要
为了提高体光照——体数据可视分析的重要手段的性能,针对现有的体光照技术存在的光照模型过于简化、只考虑光照的局部特征、收敛速度过慢的问题进行了创新研究,提出了一种基于图形处理单元(GPU)的能够支持多重散射的体光照方法。该方法在进行重要性采样时不仅根据相函数采样重要的方向,同时也考虑各方向上所携带的能量大小,因此能够更快地收敛;为了快速估算出一个方向上所携带的能量,采用了一个新颖的基于哈希桶的体数据代理,利用这个代理能够有效地跳过相似的体素,从而提高光照计算的效率。该方法相比于传统的方法能够取得更显著的加速比,同时能够有效地支持全面光照,产生复杂的光照和阴影效果。
To improve the performance of volume illumination, an important technique to visualize volume data, an effi- cient GPU-based volume illumination method supporting multiple scattering was presented based on an innovational study conducted to avoid existing volume illuminations＇ problems such as using an over-simplified lighting model, only considering local features and taking long time to converge. The new method not only considers the phase func- tion direction when doing importance sampling, but also takes the associated energy in each direction into account, so it can converge more rapidly. In order to fast estimate the energy in a direction, it adopts a novel Hash bucket- based volume data proxy, and based on this proxy, similar voxels can be efficiently skipped and the illumination cost can be reduced. The proposed method can gain significant speedup over conventional methods and can produce realistic rendering results with advanced lighting and shadowing effects.
出处
《高技术通讯》
CAS
CSCD
北大核心
2015年第10期905-911,共7页
Chinese High Technology Letters
基金
国家自然科学基金(61173067
61379085)资助项目
