Research On Anomalous Volume Modeling Method Based On Fusion Segmentation

The construction of three-dimensional models of geological anomalies is the basis for many tasks such as geospatial analysis,interpretation of geological phenomena,numerical simulation of geological processes,evaluation of mineral resources,and utilization of underground space.The construction of a three-dimensional model of geological anomalies based on three-dimensional seismic data can provide a basis for numerical simulation of oil reservoirs and deployment of reserve calculation wells.It is one of the forefront issues both at home and abroad,both academic and industrial.At present,the segmentation of geological anomalies is mostly based on single attribute segmentation.However,the single seismic attribute has multiple solutions and cannot fully and accurately reflect the anomalous body's geological structure and edge details,resulting in inaccurate segmentation results.Multi-attribute seismic data fusion is an important research topic.In the field of segmentation,the current segmentation of abnormal bodies is mostly pixel-level segmentation,and there is a problem of large amount of calculation and redundant information.This thesis starts from the two stages of the anomaly body segmentation and uses multi-attribute fusion to realize data optimization before segmentation.The segmentation algorithm is optimized in the segmentation to improve the computational efficiency and segmentation accuracy.The main work and innovations of the anomaly volume three-dimensional model based on fusion segmentation are as follows:(1)In view of the problem of insufficient description of geological anomaly for a single seismic attribute,this thesis proposes a multi-attribute fusion method based on ISOLLE.Firstly,various attributes are selected from the original amplitude data body according to the attribute category and the sensitivity to the abnormal body.Then these attributes are preprocessed.Finally,the original multiple attributes are merged into new attributes through the ISOLLE algorithm.This method considers the nonlinear relationship between seismic attribute data and is a nonlinear fusion method.The merged properties are better than the previous ones,and the accuracy of the anomaly body edge and area depiction has been improved,laying a good foundation for the next step of segmentation and reconstruction.(2)To solve the problem of too many nodes and computational complexity in the graph segmentation algorithm,this thesis proposes an abnormal body segmentation method based on super voxel and graph cut in combination with super voxels.It was first applied to the three-dimensional segmentation of geological anomalies.The method firstly generates a three-dimensional super voxel through the SLIC algorithm,and the generated super voxel fits the edge of the anomaly well and has good homogeneity.Then through a simple human-computer interaction,the final segmentation results are obtained in combination with a graph cut frame.Finally,the surface of the abnormal body is obtained by extracting the isosurface.In the actual seismic work area,the multi-seismic attributes are fused using the method proposed in this thesis.Compared with other fusion methods,the attributes obtained by the method proposed in this thesis are more accurate in characterizing abnormal body regions.In the abnormal body segmentation,compared with other traditional segmentation methods,the proposed method has been improved in both accuracy and computational efficiency.