Optimised edge detection filters in the interpretation of potential field data

Abstract

Many of the existing balanced edge detection filters of potential field data only use the feature that the vertical derivative is zero above the source edges to recognise the source edges. This will produce spurious edges in the interpretation of potential field data. In order to solve this disadvantage, a new format of the edge detection filter is presented, which produces maximum values only when the horizontal derivative is a maximum and the vertical derivative is zero, so the new filters will not produce spurious edges, and will provide more accurate results. The proposed filters are tested on synthetic potential field data. The recognised edges are shown to be consistent with the true edges, and do not produce additional edges. Moreover, the proposed filter suppresses the effect of noise, and displays the edges more clearly compared to previous edge detection filters. It is also applied to real magnetic data to obtain the horizontal locations of iron ore.