Adaptive clustering Hough transform and fault detection

Lei Chen,Chuangbai Xiao,Jing Yu,Yahong Zhang,Zhenli Wang

DOI: https://doi.org/10.3772/j.issn.1002-0470.2017.03.001

2017-01-01

Abstract:In consideration of the problem that the traditional Hough transform based methods for seismic fault detection can only detect single fault while multiple faults can not be detected accurately, a new fault detection method based on adaptive clustering Hough transform was proposed.The method is described below: Firstly, edge detection is performed on the pre-processed seismic coherent image and the Hough transform of the edge image is performed to detect line segments.Then, the line segments are clustered adaptively according to their inclination angle and position to obtain a more complete line segment.Finally, each point on the complete line segment is optimized according to the original seismic image to make the fault more accurate and smoother.To berify the effectiveness of the proposed method, the experiments were done on real seismic images.The experimental results indicated that multiple faults in seismic images were detected correctly by the proposed method.The correct rate was above 90%, and the peak signal noise ratio was improved by 10% compared with current algorithms.