A deep neural network based method for magnetic anomaly detection

Abstract Magnetic anomaly detection (MAD) is a technique to find ferromagnets hiding in strong and complicated magnetic background. In many practical cases, the targets are very far from the detection sensor, which leads to low signal‐to‐noise ratio (SNR) and high detection difficulty. Most of the c...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Yizhen Wang, Qi Han, Guanyi Zhao, Minghui Li, Dechen Zhan, Qiong Li
Formato: article
Lenguaje:EN
Publicado: Wiley 2022
Materias:
Acceso en línea:https://doaj.org/article/7246da67cecb4c0ca705c2e666166f08
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Descripción
Sumario:Abstract Magnetic anomaly detection (MAD) is a technique to find ferromagnets hiding in strong and complicated magnetic background. In many practical cases, the targets are very far from the detection sensor, which leads to low signal‐to‐noise ratio (SNR) and high detection difficulty. Most of the current methods determine the existence of target by some approaches based on signal analysis, such as the orthogonal basis function (OBF) and the minimum entropy (ME). However, although these methods consume low resources, the detection performances are not satisfactory enough. In recent years, due to the increase of computer capability, complex methods become applicable in MAD. In this study, a deep neural network (DNN) is adopted to detect the magnetic anomalies. The DNN has shown its better ability to represent natural data in many applications. A feature automatically learned by a DNN from data in the raw form is more effective for detecting target signals and suppressing irrelevant variations. Herein, a convolutional network with residual structure to implement the feature extraction is designed and an MAD method based on it is proposed. Through the semi‐real tests, the proposed method exhibits a strong capability to extract features and shows excellent performances on detection.