基于数学形态学的地震带图像识别分析

本文通过“滑动窗口”的方法将离散的地震点分布转化为反映其密度分布的灰度图像,再以数学形态学中骨架提取方法来提取地震带位置。首先对地震分布密度的灰度图像进行连通分析,认为主要的连通成分即对应了地震密集分布的地震带所处位置,之后对主要连通成分通过Fourier滤波来平滑图像内部及边缘的噪声,对经过Fourier滤波的图像运用数学形态学骨架提取方法来提取出骨架,以识别地震带的位置。本方法在使用模拟数据检验方法的有效性之后,应用于大华北地区的地震记录,对提取出的骨架采用GIS中的缓冲分析方法进行统计检验,同时结合地震专家划分的地震带进行比较,结果令人满意。

Identification of Seismic Belts from Earthquake Points by Skeleton Operation of Mathematical Morphology

This paper aims to extract the spatial distribution of discrete point events by means of image processing after conversion from the discrete field to the image field.For identifying the seismic belts,this paper creates the gray-value density image from the discrete earth-quake points by means of “sliding windows” and uses the Skeleton operation of mathematical morphology to extract out the skeleton where denotes the position of seismic belts. The identifying prodcedure analyzes the connectivity of the gray-value density image of earthquake distribution firstly. The main connective components are thought as the correspoonding position of seismic belts where the earthquakes are dense. After used the Fourier filter to smooth the main connective component image having noise,the Skeleton operation of mathematical morphology is used to extract out the skeletons as the position of identified seismic belts. After the artificial data is used to test the validity of this method,this method is applied to the earthquake records of the Great North China. The results are statistically verified by the Buffer analysis of GIS. The seismic belts drawn by seismologist are also compared. The result is satisfactory.