DeepMAD: Deep Learning for Magnetic Anomaly Detection and Denoising

Xin Xu,Xiaojun Liu,Ling Huang,Guangyou Fang

doi:10.1109/access.2020.3006795

Abstract

In this paper we introduce an end-to-end deep learning (DL) framework for magnetic anomaly detection (MAD) and denoising. This framework consists of two neural networks: a binary classification network for magnetic anomaly detection and a regression network for geomagnetic noise suppression. The two networks work in a cascade mode: the magnetic field measurement is first sent to the detection network to check the existence of the anomaly signal, and then to the denoising network for extracting the signal from the geomagnetic noise if the detection result is positive. The core idea of our proposed method is that the characteristics of both the magnetic anomaly signal and the geomagnetic noise can be learned from massive training data. The experimental results show that: (1) under the same false alarm rate constraint, the probability of detection of our proposed method is above 80% when the signal-to-noise ratio (SNR) equals -6 dB, while the orthogonal basis function (OBF) method fails when the SNR is below 0 dB; (2) for geomagnetic noise suppression, an improvement of 10 to 15 dB is achieved for data with input SNRs between -5 and 15 dB. Our results paved the way for data-driven magnetic anomaly detection and denoising.

Highlights

Magnetic Anomaly Detection (MAD) has been one of the most important methods for detecting ferromagnetic target and is widely used in submarine detection, aeromagnetic survey, etc
The orthogonal basis function (OBF) method fails when the input signal-to-noise ratio (SNR) is below 0 dB, while the CNN method still maintains a relatively high probability of detection (PD) value even for −5 dB of SNR
Our integrated MAD framework is composed of two dedicated deep CNNS, a detection network for magnetic anomaly detection and a denoising network for geomagnetic noise suppression

Summary

Introduction

Magnetic Anomaly Detection (MAD) has been one of the most important methods for detecting ferromagnetic target and is widely used in submarine detection, aeromagnetic survey, etc. There are five main sources of magnetic noise [1]: (a) Inherent sensor noise It is the upper bound on the measurement capability and generally cannot be reduced or removed except by improving the sensor or changing to a more advanced type of sensor; (b) Platform interference noise which arises from the ferromagnetic/conducting material of the platform and its rotation in the Earth’s magnetic field. Noise generated from electrical currents induced by the vertical motion of the conducting seawater, in the presence of Earth’s magnetic field It decays exponentially with altitude and can be neglected when the platform is flying relatively high [3]; (e) Geological noise which arises from the horizontal motion of MAD system across submerged concentrations of magnetic materials contained within or submerged below the seabed or ground. Since the sensitivity of advanced magnetic sensors and the performance of aeromagnetic compensation equipment have improved substantially in past decades, environmental geomagnetic noise becomes the limiting factor of the detection range for magnetic anomaly detectors

Methods

Results

Conclusion