MUSIC algorithms for rebar detection

The MUSIC (MUltiple SIgnal Classification) algorithm is employed to detect and localize an unknown number of scattering objects which are small in size as compared to the wavelength. The ensemble of objects to be detected consists of both strong and weak scatterers. This represents a scattering environment challenging for detection purposes as strong scatterers tend to mask the weak ones. Consequently, the detection of more weakly scattering objects is not always guaranteed and can be completely impaired when the noise corrupting data is of a relatively high level. To overcome this drawback, here a new technique is proposed, starting from the idea of applying a two-stage MUSIC algorithm. In the first stage strong scatterers are detected. Then, information concerning their number and location is employed in the second stage focusing only on the weak scatterers. The role of an adequate scattering model is emphasized to improve drastically detection performance in realistic scenarios.