Research On Fast 3D-Visualization Based On Large Seismic Data

In the field of seismic data processing,Geoscientists analyze and study the underground structure through seismic data and its imaging maps to obtain important information,such as the location and reserves of oil,gas and minerals.Formerly,Researchers analysis geological images only based on 2D seismic visualization because of the limitations of visualization technology.Obviously,this method of interpreting 3D structures with 2D images has some limitations,therefore,researchers began to explore 3D imaging methods of seismic data.Up to now,the 3D visualization of seismic data is still a hot research direction in the field of seismic data interpretation,its goal is to provide researchers with smooth 3D display effects in a large-scale seismic data environment while ensuring image quality,this will help the experts to understand and explain the underground structure.Direct volume rendering algorithm is one of the mainstream algorithms in 3D visualization.yet,there is a flaw when it is applied to large-scale seismic data: since the volume of the seismic data is usually larger than the size of memory of conventional computer,some direct volume rendering algorithms cannot be directly used.In response to the problem,this article has conducted in-depth study and research.In a large-scale data environment,Octree structure is usually used to re-index data and combined with LOD technology,when high-resolution data can not be loaded once or loaded too slowly,priority loading low-resolution data,then gradually loading highresolution data,Thereby achieving 3D display of large-scale data.In this paper,we have studied the above methods and tried to apply them to largescale seismic data.This paper studies the method of re-indexing seismic data by using octree.However,experiments have shown that the size of the artificially determined octree’s tile is difficult to maintain high query efficiency for seismic data of various scales,Moreover,the lack of a real-time preloading strategy in this method makes the phenomenon of frequent jamming in the continuous browsing process.Regarding the issue above,In this paper,the cluster-based Hilbert-R tree is used to re-index the Seg-Y format seismic data,which improves the indexing efficiency of the target data block,thus reducing the drawing time.At the same time,this paper proposes to use the cubic spline interpolation to fit the viewpoint trajectory in the continuous browsing process.Then predict the position of the viewpoint afterwards,and obtaining the corresponding data block by the frustum clipping method according to the predicted viewpoint coordinates.Finally,the data block is loaded into the memory in advance,thereby reducing the sense of stagnation when the geological personnel continuously view the seismic image.In order to verify the effectiveness of the proposed method,a series of comparative experiments were carried out.The experimental results show that the indexing efficiency of Hilbert-R tree based on clustering is 8.93%~14.77% higher than the traditional Hilbert-R tree,which is 23.38~40.89% higher than that of octree.Also,the cubic spline interpolation algorithm proposed by the text is compared with the Lagrange algorithm and the Hermite algorithm.It is found that the simulation effect on the motion trajectory is obviously better than the Lagrange algorithm and the Hermite algorithm.The prediction accuracy of cubic spline interpolation is 10%~16% higher than that of Lagrange interpolation algorithm,and 10%~12% higher than Hermite algorithm.Finally,it is found through experiments that even for 10 GB seismic data,the system can maintain a frame rate of about 24 frames.Therefore,the method of this paper achieves the goal of rapid display of large-scale seismic data.