基于形状描述因子的三维地震数据层面自动构建新方法研究

The energy is one of the critical factors for the development of a country and it is an important and great requirment to solve the problem of energy shortage in China. To correctly interpret the geology structure from 3D seismic data, which has important affection on the generation, movement and aggregation of petroleum and natural gas, is one of the important things in oil-gas exploration. In this project, we would present a novel concept, simplified Shape-DNA, as the shape description factor, and applied it on horizon picking and grouping in the progress of automatic construction of horizons. The main procedure of this project is: firstly, the seed points in all the orthogonal slices of the seismic data are recognized automatically, which are used for tracking the horizons in the seismic data. Then the horizons are tracked automatically based on the recognized seed points. Secondly, the tracked horizons are seperated into several groups according to the shape of the curves of each horizon. Lastly, several horizon surfaces are represented and constructed using piecewise interpolation methods. Each horizon surface is constructed from one group of horizon curves which are tracked and grouped in previous steps. The method presented here has the capability of robustly tracking and correctly grouping horizon curves suppressing the interference of faults in seismic data, due to that the simplified Shape-DNA is an operator which can describe the intrinsic features of shape. By using the simplified Shape-DNA in this project, we solve the problem of cross strata, which is one of the big problems in current methods of horizon surface construction. Meanwhile, the framework presented in this project has more degree of automation for horizon surface construction compared with those presented before. In one word, this project presented a novel workflow of automatically constructing horizon surface, which has great significance for improving the efficiency of oil-gas exploration and correctly guiding the progress of oil-gas exploration.

能源是一个国家发展的命脉，解决能源问题是我国的重大需求。油气勘探开发的一个关键问题是：如何从大量的三维地震数据中正确分析出层位面等对油气的形成、运移和聚集有重要意义的地质构造。本项目提出精简Shape-DNA作为形状描述因子，并将其用于层面构建过程中的层位线拾取和分组。首先，在全部正交剖面中自动识别出层位追踪种子点并自动追踪层位,然后,对追踪出来的层位线按照其形状相似性进行分组,最后,通过分片插值方法从每组层位线构建出一个层位曲面。由于精简Shape-DNA是一种能够对形状进行本质描述的算子，我们的方法能够排除断层的影响而对层位线进行稳定跟踪和准确分组，解决目前层位构建方法中普遍存在的容易窜层位的问题,同时,项目提出的算法框架提高了层位构建的自动化程度。本项目提出新的自动构建层位曲面的工作流程，对于提高油气勘探效率、正确指导油气勘探开发过程有重要意义。

能源是一个国家发展的命脉，解决能源问题是我国的重大需求。油气勘探开发 的一个关键问题是: 如何从大量的三维地震数据中正确分析出层位面等对油气的形成、运移和聚集有重要意义的地质构造。本项目提出精简Shape-DNA 作为形状描述因子，并将其用于层面构建过程中的层位线拾取和分组，最终实现自动的地震层位面构建。.本项目的研究目标是解决如何从大量的三维地震数据中正确分析出层位面结构的问题，课题组研发出了紧凑型Shape-DNA作为网格形状的描述算子，并将其用于层面构建过程中的层位线拾取和分组问题，结合课题组自定义的八叉树数据结构，可快速的对三维地震层位面进行检测和提取，并提出新的自动构建层位曲面的工作流程，取得了满意的结果。在保证对三维形状提供较高辨识度的前提下，紧奏型Shape-DNA算子仅需要10个浮点数既可以对任意三维形状进行精确的描述，大大减少了描述三维形状所需要的存储空间，从而显著减少进行三维形状检索及匹配的计算时间，是本项目研究成果的重要基础。项目组还开展了大规模三维地震数据的快速可视化研究，基于项目组自定义的八叉树数据结构，设计了一种分辨率自适应的多分辨率大规模数据可视化流程，结合数据加载、预加载技术，可实现超大规模数据三维可视化的实时渲染和交互。项目组还研究了水下自主机器人定位和无线传感网连通集最大化问题，提出的算法可有效增加网络整体生存时间，定位也更准确和安全。进一步，项目组研究了同样具有分层结构的视网膜层析图像的自动分层问题，基于连续最大流算法，课题组提出了一种新的图像自动分割算法，与手工分层的结果相比，该算法能够更加准确的对视网膜图像进行分层。.由于精简Shape-DNA是一种能够对形状进行本质描述的算子，我们的方法能够排除断层的影响而对层位线进行稳定跟踪和准确分组，解决目前层位构建方法中普遍存在的容易窜层位的问题,同时,项目提出的算法框架提高了层位构建的自动化程度。本项目提出新的自动构建层位曲面的工作流程，对于提高油气勘探效率、 正确指导油气勘探开发过程有重要意义。