Automatic Fault Surfaces Extraction from Seismic Volume

Abstract

Fault extraction is an important part of understanding geologic structures in 3D seismic images. Manual fault interpretation is very time consuming so automation of this process is very desirable. Snakes/active contours is an versatile tool in computer vision for finding contours. In this paper we have implemented two variants of Snakes based algorithms. One extension of the classic Snakes algorithm with an shape prior element. And one implementation of a basic spline based Snake, B-snake. A fault attribute has also been implemented, and a localized version of the fault attribute has been combined with the B-snake model and together they have shown some very promising results with extraction of faults.