I. Introduction
Faults are an important geotectonic feature formed by the brittle deformation of subsurface rock formations under the action of compressional or tensile forces [1]. Fault detection and interpretation are essential in understanding the underground structure and reservoir locations from seismic profiles and are crucial to geological hazard prevention and control. In seismic migration images, faults are usually depicted as lateral discontinuities. Traditional fault detection methods analyze and calculate the properties of seismic migration images to find these discontinuities. Several methods for fault identification have been proposed in previous investigations, including similarity [2], coherence [2], variance [4], and curvature [5]. In the application of complex geological tectonic zones, however, the accuracy and dependability of these methods are still problematic.