应用句法识别实现地球物理磁异常的自动划分

AUTOMATIC DIVIDING OF MAGNETIC ANORMALY BY STRUCTURAL RECOGNITION

地球物理磁数据中蕴含着丰富的数值信息和结构信息综合运用数值信息中的统计特征和结构信息，并结合地质和实际情况是磁异常解释的一条重要途径本文在前人工作的基础上〔１〕，尝试应用句法模式识别和聚类分析的方法，根据磁异常的结构和统计特征对磁异常进行图形化、机器自动地划分，为高精度、快速和可靠的磁测解释提供一种手段本文首先用句法模式识别方法对磁异常进行结构划分，找出基节点，提取基元，然后用“衍生树”〔２〕的方法形成基类；最后用聚类分析方法将各基类按距离相似性规则进行归类，从而达到磁异常分类解释的目的文章在理论分析和正演模拟的基础上，验证了方法的可行性；进一步结合三峡某地考古的实测磁数据证实了方法的实用性；取得了较为明显的效果，为模式识别技术在地球物理综合解释中的应用开辟了一条新路子。

There are a lot of digital and structural information stored in the geophysical data It will be much favorable if we can utilize all these information,combined with the known geological information and the field information,when we process and interpret magnetic data In this paper,we introduce Structural Recognition and Cluster Analysis method to label the magnetic anomalies automatically and visually,based on the works which has been done 1 ,so that we can provide a computer assisted tool for accurate interpretation of magnetic data First we employed the structural recognition method to label the magnetic anomalies into basic nodes,which then clustered into groups,we called them basic units Secondly,we combined certain units into a class,according to the concept ″generating tree″ 2 ,which was called the basic class Finally,we classified these basic classes with respect to the rule of distance resembility through cluster analyzing The method has been proved to be reliable and efficient since we applied it to modeling and field data It was believed that the method would open a new way to integrated interpretation of geophysical data,and founded a base for later study