Target Detection for Low Signal-to-Noise Ratio Scalar Magnetic Unexploded Ordnance Surveys: A Multilevel Orthogonal Basis Function Approach

With the increasing speed of magnetic data acquisition by uncrewed platforms, unexploded ordnance (UXO) surveys now face challenges such as susceptibility to environmental noise interference and low data acquisition. This study proposes a multilevel orthogonal basis function (MOBF) detection method to address the challenges of weak magnetic anomaly detection (MAD) in complex noise environments, particularly for UXO surveys. The MOBF method integrates discrete stationary wavelet transform (DSWT) and 2-D orthogonal basis function (2D-OBF) processing through a cascaded decomposition-fusion architecture. By leveraging DSWT’s shift-invariant multiscale decomposition, the method effectively separates colored noise (with a power spectral density (PSD) of 1/<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${f}^{\,\alpha }$ </tex-math></inline-formula>) from target signals, while OBF enhances localized spatial correlations of anomalies. A variance-weighted energy fusion strategy is introduced to aggregate multiresolution features, significantly improving signal-to-noise ratio (SNR). Numerical simulations demonstrate MOBF’s robustness across diverse noise scenarios: at −20 dB SNR under Gaussian noise, the MOBF method has a higher detection probability and lower false alarm rate than traditional methods. In colored noise environments, MOBF maintains reliable detection at −15 dB SNR, whereas 2D-OBF fails. Field tests conducted in coastal areas with uncrewed aerial vehicle (UAV)-borne magnetic surveys validate MOBF’s practicality, successfully identifying ferromagnetic targets (anchors, iron tools) under challenging conditions (strip noise). Despite limitations in distinguishing UXOs from nonhazardous ferromagnetic objects, MOBF exhibits superior noise immunity and spatial resolution compared to existing methods. The proposed method provides a viable solution for real-time UXO detection on mobile platforms, particularly in low SNR scenarios with colored noise interference.