Seismic Fault Analysis Using Graph Signal Regularization
Patitapaban Palo, Aurobinda Routray, Sanjai Kumar Singh · 2021 29th European Signal Processing Conference (EUSIPCO) · 2021
Graph signal processing (GSP) is a process that deals with the processing of data that graphs can support. The concepts of digital signal processing (DSP) are mapped to define the processing techniques for signals on a graph. Furthermore, a seismic fault is nothing but a discontinuity in a volume of seismic data across which there has been significant seismic displacement. In this paper, we propose an approach to interpret the presence of a fault in seismic data using GSP. First, we extract 2D patches of data from a seismic volume and present the seismic data in the graph domain by considering a seismic patch as a vector space node. K-nearest neighbour graph is used to present data on graphs. Next, we do the fault classification by implementing a graph signal regularization technique that uses the graph shift matrix.