三维地震数据场可视化并行算法研究

【作者】 章纪扬；

【导师】 张二华；

【作者基本信息】 南京理工大学 ， 计算机应用技术， 2008， 硕士

【摘要】 三维地震数据场的可视化对石油勘探起着重要的作用,它把采集到的物理数据转换成图形、图像,并通过给体数据赋颜色和不透明度,以及改变观察角度等方式来展现三维地震数据场的各种特征。而直接体绘制技术是三维数据场可视化的主要技术手段之一,它不仅能够表现出三维对象的表面特征,也能显示其内部信息,因此被广泛地应用于三维地震数据场的可视化中。然而,直接体绘制技术却有着计算量大、计算时间长的缺点。实际生产过程中的三维地震数据场的数据量往往十分巨大,目前即使应用高性能的单台计算机,在对三维地震数据进行交互式可视化过程中也难以达到实时显示的效果。因此,研究大规模三维地震数据的并行可视化技术显得尤为重要。本文基于直接体绘制中的光线投射算法,首先分析了该算法的原理,设计了三维地震数据可视化的程序,在分析了影响并行算法性能的多种因素后,结合具体的计算机集群设计了一套并行光线投射算法,并用实际三维地震数据对该算法进行了测试试验;最后,利用二叉树的特点对算法进行改进,以减少子图像的合成步骤。与串行算法相比,并行光线投射算法的绘制速度得到了很大的提高。更多还原

【Abstract】 Visualization of 3D seismic data field plays an important role in oil prospecting. It transforms physical data collected into graphics and images, and shows the features of the 3D seismic data, by assigning color and opacity to volume data, and changing the viewing angle. Direct volume rendering technology is one of the main methods for 3D visualization. It can show not only the external features of the 3D objects, but also the inner information. So, it is applied widely for visualization of 3D seismic data filed. But volume rendering has a shortcoming, which is that it is often too computationally intensive.The amount of the data in a 3D seismic field is usually huge in our practical production, so that the computer with the best performance can’t have a real time effect when doing interactive visualization of 3D seismic data at present. So, it is necessary to research on parallel visualization technology of large-scale 3D seismic data.Our paper is based on ray casting algorithm of direct volume rendering. First, ray casting algorithm is analyzed and realized. Then, after analyzing the factors which affect the performance of parallel algorithm, a parallel ray casting algorithm is designed and realized on a computer cluster for testing with 3D seismic data. Finally, the algorithm is improved, in use of the characteristic of binomial tree, to reduce steps of the sub-images’ synthesis. Compared with the serial algorithm, the rendering speed of parallel ray casting algorithm is increased greatly.更多还原