位场数据边界识别的新方法—增强型水平导数法  

New edge detection method of potential field data-enhanced horizontal derivative method

作　　者：马国庆[1] 杜晓娟[1] 李丽丽[1] 

机构地区：[1]吉林大学地球探测科学与技术学院,长春130021

出　　处：《地球物理学进展》2013年第1期402-408,共7页Progress in Geophysics

基　　金：深部探测技术与实验研究专项项目SinoProbe-09-01(201011078)

摘　　要：边界识别是位场数据解释的一项基本任务.现有边界识别方法大多是基于水平与垂直导数的高通滤波器,但垂直导数的计算会明显地增大噪声的干扰.为了改善这一问题,本文提出了一种新的边界识别滤波器,该方法是利用不同阶水平导数之间的线性组合来进行地质体边界的识别,由于该方法不需要垂直导数参与计算,因此其输出结果较稳定.理论模型试验表明,该方法能够更加清晰的识别地质体的边界,且与其真实边界相吻合.最后将其应用于实际数据的处理中,其结果清晰地反映出断裂的位置及构造之间的界线,并能识别出更多的细节信息.Edge detection is a common task in the interpretation of potential field data. There are many high-pass filters based on horizontal derivative and vertical derivative in use, but the computation of vertical derivative can significantly enhance the effect of noise. In order to improve this problem, we present an edge detection filter, which use the various combinations of different order horizontal derivatives to recognize the edges of the sources. The new filters don＇t require the computation of vertical derivatives, which as well as making the computation more stable. The new filters are demonstrated on synthetic gravity anomalies, the new filters can display the edges of the sources more precisely and clearly. At last, we applied the new filters to real gravity and magnetic data, and the results can display the horizontal locations of the structures and the boundaries of the stratums, and can recognize more subtle details.

关 键 词：边界识别 位场 水平导数 细节 

分 类 号：P631[天文地球—地质矿产勘探]