A Method Of Fault Surface Automatic Extraction From Three-dimensional Seismic Data

The high-speed development of society has brought about huge progress of various productive and processing techniques, which putting forward higher request to model reconstruction and repairing techniques of industrial production. And as the most visualized appearance of physical model, three-dimensional geometric model gets the favor of most researchers. Therefore the research of three-dimensional surface reconstruction techniques has become a research hotspot. The same thing happens in the field of seismic exploration, in which geological researchers hope to be able to use three-dimensional volume rendering technology to display geological structure comprehensively. In the field of seismic exploration, seismic data interpretation plays a crucial role, and the fault interpretation is one of the core parts in seismic data interpretation.Conventional fault interpretation methods mostly adopt the method of manual interpretation, in which geological interpreters analyze faults in the vertical sections or horizontal sections of three-dimensional seismic data relying on their abundant geological knowledge and experience. This method is long-cycle, subjective and difficult. In order to reduce the cycle of the explanation and overcome the subjectivity of fault interpretation, a growing number of researchers hope to be able to realize automatic or semi-automatic fault interpretation with the help of coherence cube.Taking fault extraction as an example, this paper put forward a set of fault surface automatic extraction process from three dimensional data. The method dealt with faults globally in three dimensional spaces with the help of coherence cube. Moreover, this paper took the method of three dimensional point cloud surface reconstruction to reconstruct fault surfaces. The followings are the main work of this paper:1. We put forward a new process to extracted faults from the existing ant cube. The process dealt with faults globally in three dimensional spaces with the help of image segmentation technology, three-dimensional image processing technology, data clustering analysis theory and the spatial surface reconstruction technology from point cloud. Firstly we executed a binaryzation process to separate fault points from background points. Then we denoise the binary images to solve the problem of sticking problem between faults. Finally with the help of surface reconstruction technology we reconstructed three-dimensional fault planes from point cloud.2. Based on the binary data after pretreatment, this paper proposed a method which using the idea of three-dimensional point cloud surface reconstruction to build fault surface. This method extracted fault surfaces through the process of scatter through clustering, fault re-division and surface reconstruction.3. Considering the truth that the survey areas for interpretation are usually big, and the memory of computer today is very limited compared with the huge geological data. So this paper also designed a feasible data partitioning algorithm to deal with huge geological data with the limited computer memory. Taking hard disks of computer as buffers, we partitioned the geological data into several equal-size parts and kept them in computer’s hard disks firstly, and then we loaded corresponding data blocks into memory for processing based on needs. Thus we can solve the problem of insufficient computer memory.This paper applied the proposed algorithm to F3 survey area in the end. It turns out that the proposed algorithm can extract the fault planes which implied in the data effectively when comparing with Petrel. And taking advantage of the proposed data partitioning scheme, we can make good use of limited memory resources to solve the problem of fault extraction from big data.