A high-SNR ultrasonic imaging method for weakly heterogeneous medium

Ultrasonic full matrix capture (FMC) dataset has been widely used in various fields including biomedical imaging, non-destructive detection and structural health monitoring. However, the measurement medium is usually weakly heterogeneous, which leads to the speckle artifacts in the imaging result of the conventional methods and reduces the signal-to-noise ratio (SNR). To overcome this deficiency, this paper proposes a novel high-SNR method to recover the targets in the weakly heterogeneous medium clearly from the FMC dataset. Singular value decomposition (SVD) is utilized to extract the signal subspace, and it transforms the passive reflector detection problem into an active source location problem. Then, the beamforming is conducted with the cross-spectral matrix calculated by the signal subspace. Furthermore, the deconvolution process is implemented with the non-stationary point spread function (PSF) for the further suppression of speckle artifacts. In addition, the phase shift factor is introduced to make the proposed method adapt to the multilayer medium. Three existing imaging methods for FMC datasets are introduced to make comparisons with the proposed method. The simulation and experiment results demonstrate that the proposed method can recover the image with much higher SNR (>20 dB) wherein the speckle artifacts in the background are well suppressed. Additionally, the proposed method is more efficient than total focusing method, and it only takes about 2 s to obtain the high-quality imaging result.