层位面追踪是三维地震数据解释的一项关键操作,对于地质勘探和油藏工程有重要的指导意义.传统的层位面自动追踪算法采用的是种子点蔓延法,该方法是一个不断进行迭代计算的过程.由于三维地震数据的数据量很大,传统的基于CPU的层位面自动追踪算法的计算效率不高.文中提出了一种基于CUDA的三维层位面自动追踪算法.该算法对种子点蔓延法进行了改进,实现了层位面自动追踪的GPU并行解决方案.实验表明,与传统的基于CPU的层位面自动追踪算法相比,本算法的计算效率有明显的提升.%Horizon tracking is an important operation in 3-D seismic data processing, and the tracked horizon can be used by the experts to detect the earth's subsurface structure. Processing speed is a major bottleneck in seismic horizon tracking since the data size of 3-D seismic data is usually very huge. In this paper,present a new automatic horizon tracking algorithm, which is based on CUDA technique. The algorithm fully utilizes the many-core structure and the high performance computing ability of GPU, realizing a large scale parallel method for horizon tracking. The results of experiments show that the algorithm has an obvious speedup than the traditional CPU-based horizon tracking algorithm.
展开▼