【作者】 王文龙；

【导师】 邢建国；

【作者基本信息】 浙江工商大学 ， 计算机应用技术， 2010， 硕士

【摘要】 体绘制技术是三维数据场可视化技术的一个重要分支,在医学、地质勘探、流体力学等领域均有广泛应用。研究者们提出了多种体绘制算法,其中光线投射算法是现有体绘制算法中图像效果较好的一种,但是由于投射光线数据量巨大,对每条光线又需多次采样,导致生成图像的速度较慢。目前体绘制方法的实现主要有两种：软件实现和硬件实现。软件实现在体绘制中应用得十分广泛,但绘制的体数据往往比较庞大,只依靠软件加速绘制尤其是绘制具有整体光照效果的图像时,算法性能还是有限的。因此人们将目光转向图形硬件,希望借助硬件提供的相关功能加速体绘制。虽然目前的这些专用体绘制硬件系统能够达到一定规模数据场实时体绘制的要求,且算法流水线根据硬件结构经过了最优处理,实时绘制性能得到提高,但算法的流程固定不变,其灵活性相对比较差,而且对硬件配置要求很高,价格昂贵。本文采用可配置的FPGA解决方案,提出了一种基于软硬件协同设计的高效并行光线投射算法框架,把计算复杂但计算量少的任务和人机交互的任务通过软件实现,而那些计算简单且计算密集型的任务通过硬件并行实现。这样既提高了绘制速度,又兼具友好的交互性,具有更好实用性。直方图引导的自适应局部传递函数方案的提出保证了高质量图像的生成,而且增强了图像的细节信息。算法框架的提出使便携式小型化的诊断系统的实现成为可能。更多还原

【Abstract】 Volume rendering is a significant part of 3D data visualization, is widely used in medicine, geological survey, hydrodynamics, and so on. Researchers propose several volume rendering algorithms, in which Ray Casting is more efficient, but enormous rays and sampling lead to longer time.From now on, there are mainly two implementations of volume rendering, one is software implementation, the other is hardware implementation. Software implementation is widely used in volume rendering, but enormous volume data results in worse performance. Thereafter researchers tend to graphic hardware to improve rendering performance. These special hardware rendering systems can achieve real-time requirement, and algorithm pipelines can be optimized, yet algorithm flow is fixed, the flexibility is bad, requirement of hardware configuration is great, and the price is expensive.This paper uses configurable FPGA resolution, and proposes an efficient parallel Ray Casting based on hardware/software co-design, which implements complicated but less computational jobs and Human-Computer interaction by software, and implements simple but computation-intensive jobs parallel by hardware. Thus rendering speed is improved, and it has good HC interaction and practicability. Proposal of histogram guided adaptive local transfer function can improve the quality of rendering image and enhance details of image. The proposal of algorithm framework makes implementation of portable mini diagnose system possible.更多还原