论文部分内容阅读
本文提出一种根据信号特征划分地震图象的基于规则的解释系统。该系统由两部分组成,一个结构分析器和一个智能解释器。结构分析器采用由劳斯研究出的结构能量测定方法,从织构状信号图象中提取判别特征。分析器的主要功能是根据提取的特征测量值给图象的每个结构单元配一个初始确定度因子(CF)向量。CF 向量的元素基本上对应于该结构单元隶属度,即该结构单元隶属于图象中各结构区域的程度。智能解释器是一个基于规则的系统,由一个知识库,一个推理机,和一个并行的区域扩展控制器组成。只要一个扩展区域请求对其边界结构单元之一进行分类,就会形成一个含有该结构单元及其相邻单元有关信息的事实表。解释器得到此事实表后,即在知识库中搜索寻找可执行的规则。在执行一系列的规则并对 CF 向量进行运算后,形成一个最终 CF 向量。若此向量的对应元素超过一给定门限值,而使该向量偏向请求区域,则将此结构单元划归这一结构类,并将其并入请求区域。曾用一段地震剖面对该系统进行试验运行。结果表明,该系统能非常成功地将其划分为具有共同信号特征的多个区域。比较其它常规划分技术而言,该智能解释机的一个突出的优点是:对于原始数据中似乎未能提供足够的判别特征信息的小块图象,它具有补缀功能。本文还提供了关于墨西哥湾实际地震数据试运结果的实例。
This paper presents a rule-based interpretation system for seismic image segmentation based on signal characteristics. The system consists of two parts, a structural analyzer and an intelligent interpreter. The structure analyzer uses the structural energy measurement method developed by Rolls to extract the discriminant features from the texture signal images. The main function of the analyzer is to assign an initial deterministic factor (CF) vector to each structural element of the image based on the extracted feature measurements. The elements of the CF vector basically correspond to the degree of membership of the structural unit, ie, the degree to which the structural unit belongs to each structural area in the image. A smart interpreter is a rules-based system that consists of a knowledge base, an inference engine, and a parallel zone expansion controller. As soon as an extended area requests the classification of one of its boundary building blocks, a fact table containing information about the building block and its neighbors is formed. After the interpreter gets this fact table, it searches through the knowledge base for executable rules. After executing a series of rules and computing CF vectors, a final CF vector is formed. If the corresponding element of this vector exceeds a given threshold and the vector is biased towards the request area, the unit is assigned to this structure class and incorporated into the request area. The system has been experimentally operated with a seismic section. The results show that the system can be very successfully partitioned into multiple regions with common signal characteristics. An outstanding advantage of this intelligent interpreter compared to other conventional partitioning techniques is that it has a patching function for patch images in the original data that seem to fail to provide sufficient discriminative feature information. This article also provides an example of the results of the commissioning of actual seismic data in the Gulf of Mexico.