The Parallel Visualization Technology Research Based On Large Seismic Data

Three-dimensional seismic exploration has become today’s an important means of the exploration and development of oil and gas. The visualization of3D seismic data field plays an important role in the exploration of oil, which transforms the collected data into graphs and images, and shows the features of the3D seismic data by changing the viewing angle to improve the efficiency of oil and gas exploration. The3D seismic visualization once done by supercomputer or professional graphics workstation was very expensive and inconvenient to use. So in order to make full use of the powerful parallel calculating ability of GPU, and explain the seismic data on an ordinary PC without using expensive supercomputer or workstation, this paper brings up the CPU and GPU working together in parallel volume visualization plan---The Parallel Visualization Technology Research based on Large Seismic Data.It is difficult for dynamic display because of the direct volume visualization techniques’ calculation of large and long computing time. In order to achieve the practical application of direct volume visualization technology, in this article, texture mapping techniques are used to realize the visualization of3D seismic data, which can greatly increase the body display speed, and improve the quality of images generated at the same time. The core ideas of the algorithm in this article are as follows:first is the adaptive decomposition of original3D seismic data, so that the data can be loaded into memory; then the data in memory breaks adaptively up secondly according to the parallel ability of GPU(this means the quantity of GPU),and the data of decomposition are assigned to the kernel for parallel computing at the same time; finally we can get the final displayed results of3D seismic data with the fusion of calculated data.The main focuses of this paper are as follows:(1) the research large-scale3D seismic data’s visibility test algorithm. The large-scale3D seismic data are under three dimensional space decomposition according to its spatial distribution characteristics. After the decomposition, we can begin the visibility test, the empty space jump and sub-block ahead of deadline is the operation of each sub-block can be removed empty block and the block sub-block; early opacity cut-off based on each pixel carry out the operation, when the opacity reaches a preset threshold, termination on the back of the body elements are drawn.(2) segy format to gsegy format. gsegy file is generated specifically for the visualization of large-scale data, which puts the large-scale data into re-organization and management according to certain rules for the propose of increasing the efficiency of all kinds of2D and3D visualization. So, in this paper gsegy files are used to achieve the efficient3D visualization of seismic data. Different from the segy files’road sequence storage, the seismic data in gsegy files are re-organized by the structure of the octree. Octree structure (Octree) is a tree-level data structures used to describe three-dimensional space. The structure of segy data is a spatial three-dimensional grid structure, so the process of transforming segy file into gsegy file is a conversion process from grid structure to octree structure.(3) the research of parallel volume rendering algorithm based on the coordination of the GPU and CPU. First is the adaptive decomposition of original3D seismic data, so that the data can be loaded into memory; then the data in memory breaks adaptively up secondly according to the parallel ability of GPU (this means the quantity of GPU), and the data of decomposition are assigned to the kernel for parallel computing at the same time; finally we can get the final displayed results of3D seismic data with the fusion of calculated data.In this paper, we study data1-18.4M、data2-41.6M、data3-430.6M、data4-2623.6M、data5-10200M, five different sizes of3D seismic data, by format conversion and GPU parallel speedup, the load time of the3D seismic data and the program’s running time has improved to some extent. Compared with the direct volume visualization algorithms based on software and the volume visualization algorithm without visibility test based on hardware, the proposed algorithm in this paper has an obvious advantage, the display speed of the algorithm is based on the software direct volume visualization algorithm about10times.Experimental results show that the parallel visualization algorithm based on3D texture mapping, we can greatly improve the display speed of3D seismic data and improve the processing efficiency of the entire system, for the earthquake personnel to explain the geological structure provides a convenient, and save a tremendous earthquake staff time.