Fast Non-Stationary Deconvolution of Ultrasonic Beamformed Images for Nondestructive Testing

This paper addresses high-resolution ultrasonic image reconstruction from Full Matrix Capture (FMC) data in the context of nondestructive testing (NDT). In order to reduce numerical complexity, time-domain and model are projected into domain through a linear beamforming procedure. The resulting is interpreted as shift-variant convolution process, affected by non-stationary colored noise. An interpolation procedure built account for spatial variations point spread function. Under same methodological framework, an approximate whitening filter proposed incorporated forward model. Both constructions then allow fast computations limited memory storage. Deconvolution performed minimizing least-squares misfit error, with penalization term favoring sparsity continuity output images. Results synthetic show that approach gives performances close inversion raw FMC data, while being computationally much more efficient. method finally applied laboratory inspection stainless steel block containing closely spaced small side-drilled holes. Successful detection separation achieved flaws diameters six times smaller than wavelength, distant each other three less resolution limit given Rayleigh criterion.