Wideband Automatic Diagonal-Loading Beamformers for Acoustic Imaging

The data-independent delay-and-sum beamformer (DAS) is widely used for acoustic imaging applications, although it has lower resolution and worse interference rejection capability than the standard Capon beamformer (SCB). However, the performance of SCB can be degraded by many factors existed in practice, for example, array covariance matrix estimation errors due to a small sample size. In this paper, we have applied automatic diagonal loading method to improve the robustness of SCB. Different with other existing robust approaches, the diagonal loading parameters of the method are fully automatically chosen by either a ridge regression (RR)-based approach or a shrinkage-based approach. As shown in the numerical examples, the proposed methods perform better to estimate the sound pressure level (SPL) of the signal of interest (SOI) for acoustic imaging applications in a small sample size than DAS and SCB.