Feature removal and isolation in potential field data

Abstract

With the aim of designing signal processing tools that act locally in space upon specific features of a signal, we compare two algorithms to remove or isolate individual anomalies in potential field profiles. The first methods, based on multiscale edge analysis, leaves other features in the signal relatively untouched. A second method, based on iterative lateral continuations and subtraction of anomalies, accounts for the influence of adjacement anomalies on one another. This allows a potential field profile to be transformed into a number of single anomaly signals. Each single anomaly can then be individually processed, which considerably simplifies applications such as inversion and signal processing. © 2004 Commonwealth Scientific & Industrial Research Organisation.