A Seismic Fault Recognition Method based on Linear Regression

Lei CHEN,Chuang-Bai XIAO,YU Jing

2017-01-01

Abstract:Fault recognition is an important section in seismic interpretation region and there have been many methods for this technology, but no one can recognize the fault exactly enough. For this problem, we proposed a new fault recognition method based on linear regression which can locate the position of a fault precisely and then extract it from the seismic section. First, the seismic horizons were labeled out by the eight adjacent connected component labeling method as connected regions; second, the horizontal endpoints of each connected component were found out based on the column coordinate of the pixels within the component; finally, the linear regression method based on least sum of square error was used to generate the direct line which was used to fit the horizontal endpoints. As a result, the direct line was regarded as the desired fault. To validate the availability and advancement of the proposed method, different fault recognition methods were compared through experiments on the synthetic seismic model data and the real seismic data. The comparison of the fault recognition results indicated that the proposed method is more accurate and effective than the traditional and latest presented methods.