Characterisation of small embedded two-dimensional defects using multi-view Total Focusing Method imaging algorithm

Abstract Defect characterisation is an essential step of nondestructive testing which aims to determine the shape and the size of a defect once detected. The imaging algorithm multi-view Total Focusing Method (TFM) has been successfully used to accurately size defects of size greater than several times the wavelength. However, characterising smaller defects remains difficult due to the diffraction limit. This paper introduces as a proof of concept a probabilistic characterisation technique where the unknown defect is compared against a database of reference scatterers obtained with an ultrasonic model. The measured TFM amplitudes act as a defect signature. A Bayesian framework is used to retrieve the reference defects which most credibly explain the measurements, and to calculate uncertainties.