A novel W-MUSIC algorithm for GPR target detection in noisy and distorted signals

Abstract

When applied to ground penetrating radar (GPR), the multiple signal classification (MUSIC) algorithm is an important frequency estimation method as it can detect very closely spaced targets, particularly when one of the target responses is substantially less than another. The MUSIC algorithm however must be seeded with the number of targets to find and will indicate that number of targets regardless of the number of targets actually present. In addition, there is no reliable indication of the magnitude of the responses of the targets. In the presence of relatively low levels of noise the MUSIC algorithm is prone to reporting the position of false target positions in preference to weaker genuine target responses. In this paper a superimposed MUSIC technique is proposed to suppress false targets. A novel windowed FFT MUSIC (W-MUSIC) algorithm is examined using a linear sweep frequency in noise, and it is seen to give a clear indication of targets. In the presence of distortion, FFT is seen to be a more robust approach than the MUSIC algorithm, the latter method becoming very prone to inaccuracy in the presence of distortion.