Full Vector Inversion of Magnetic Microscopy Images Using Euler Deconvolution as Prior Information

Abstract

Abstract Paleomagnetic data is collected from bulk samples, containing a mixture of stable and unstable magnetic particles. Recently, magnetic microscopy techniques have allowed the examination of individual magnetic grains. However, accurately determining the magnetic moments of these grains is difficult and time‐consuming due to the inherent ambiguity of the data and the large number of grains in each image. Here we introduce a fast and semi‐automated algorithm that estimates the position and magnetization of dipolar sources solely based on the magnetic microscopy data. The algorithm follows a three‐step process: (a) employ image processing techniques to identify and isolate data windows for each magnetic source; (b) use Euler Deconvolution to estimate the position of each source; (c) solve a linear inverse problem to estimate the dipole moment of each source. To validate the algorithm, we conducted synthetic data tests, including varying particle concentrations and non‐dipolarity. The tests show that our method is able to accurately recover the position and dipole moment of particles that are at least 15 μm apart for a source‐sensor separation of 5 μm. For grain concentrations of 6,250 grains/mm3, our method is able to detect over 60% of the particles present in the data. We applied the method to real data of a speleothem sample, where it accurately retrieved the expected directions induced in the sample. The semi‐automated nature of our algorithm, combined with its low processing cost and ability to determine the magnetic moments of numerous particles, represents a significant advancement in facilitating paleomagnetic applications of magnetic microscopy.

Keywords

• magnetic microscopy
• geophysics
• Python
• inverse problems
• image processing