Optimization of scalar magnetic gradiometer signal processing

The work is devoted to the detection of hidden ferromagnetic objects by employing a gradiometer comprising two scalar magnetic sensors. The hidden object is modeled by a magnetostatic dipole. Processing of the gradiometer signal is carried out by evaluation of the signal energy in the space of four orthonormal functions. This procedure is implemented within a moving window covering certain number of successive samples measured equidistantly along the survey track. This work proves that the width of the moving window (or the number of the samples within the window) can be optimized so to provide maximum signal-to-noise ratio. Removal of constant bias and linear temporal trend, usually accompanying the observed survey signal, has been proved as an important step in the data processing. The latter removal relies on linear regression procedure, and the related window width is also optimized. It is shown that the signal processing should depend on several survey channels operating simultaneously. Relying on the present work, a proper number of channels for multi-channel detection algorithm can be easily determined for each magnetic search scenario.