Research On The Extraction Of Texture Properties Of Multi-Dimensional Images

Texture attribute extraction of seismic data is an important method for the seismic data interpretation.Among all the methods for extracting texture attributes,the coherence algorithm is the most representative one.It can extract faults by highlighting discontinuities in seismic data and improve the efficiency of fault interpretation.Since the fault plays an extremely important role in the exploration and development of oil and gas,the development of finer coherence algorithms and fault enhancement techniques to achieve accurate characterization of faults has been a hot and difficult issue in geological research.In this thesis,we first discusses the significance of extracting texture attributes from seismic data,and illustrates the important role of coherence technology in fault detection in oil and gas exploration;Then it discusses the current domestic and international research status of coherence technology and fault enhancement technology;the relevant principles of the texture attribute extraction method involved in this paper include the characteristics analysis of seismic data,the principle of coherent body technology,and the basic knowledge of tensor and the principle of compressive sensing.Finally,this paper proposes a solution based on the existing problems in the existing technology.T-SVD wide azimuth coherence algorithm,Ho-RPCA based fault enhancement method.The detailed work includes the following::1.For the existing wide-azimuth coherence algorithm,the coherent values are usually calculated after wide-azimuth superposition,which can lead to the loss of some high-dimensional information.This thesis improves on the C3.5 coherence algorithm and introduces the T-SVD tensor decomposition to expand the coherence algorithm to four-dimensional tensor for calculation.The improved method makes full use of the internal characteristics of high-dimensional data with wide azimuth,and it can predict small-scale faults well,providing a new idea and method for fracture detection and reservoir prediction.2.This thesis proposes a fault enhancement method based on Ho-RPCA.This method establishes the low-rank tensor recovery model to separate the low-rank fault portion and the sparse noise portion of the coherent volume data,thereby achieving fault enhancement.First,a new objective function is constructed according to the model;then the optimal solution of the variable is solved by the alternating direction method of multipliers.This method can effectively preserve the obvious fault structure,enhance the small faults in the background,and remove the extraneous noise at the same time,which can achieve ideal results.Finally,the above-mentioned wide-azimuth coherence algorithm and fault enhancement algorithm are applied on the seismic data volume of several real work areas.The results show that the practical effect of this method has certain advantages compared with the traditional method,and can achieve better fault enhancement effect.