2D Acoustic Source Localization with a 1D Array and MVDR

Hi, I am working on a MVDR ultrasound beamformer which localizes acoustic scatterers in 2 dimensions, the axial and lateral. Currently, I am passing data from a 64-element array to a MVDR beamformer one lateral position at a time, such that the input data is length_axial_dim * num_elements. This produces excellent results in improving axial resolution. When I tried a similar approach in the lateral dimension, the results are not as good, probably because the axial sampling is much better than lateral sampling to begin with. However I am wondering if there is a way to minimize the variance in 2 dimensions simultaneously?

I have looked into the 2D MVDREstimator, but that requires a 2D array. Does it make sense to model the time of flight as a second array dimension? Or is there a better way to do this? Any suggestions or comments would be appreciated thanks!

Answer 1

[William Rose]

@Jad El Harake,

I think this question is a matter for investigation, and that you will have to experiment. The two dimensions in the estimator phased.MVDREstimator2D() are two angles. In your case, the dimensions are lateral angle and axial depth. Therefore, phased.MVDREstimator2D() may not do what you want. But there is no harm in trying - except time lost if it doesn't work.

This paper is interesting and may be useful.

This paper is also of interest.

Good luck.