Improving the resolution of ultrasonic data through deconvolution

Ultrasonic nondestructive testing data contains information about breaks in the inspected medium. Discontinuities due to a defect can be identified through the interpretation of those data. Still, the interesting information about the discontinuities does not always appear directly in the ultrasonic data. Firstly, the presence of each discontinuities is “diluted” in the data; such an effect creates difficulty for the data interpretation, especially when the discontinuities are very close together, several echoes are then mixed in on widely spread echo. Secondly, when the structure of the medium granular, the ultrasonic data can be corrupted by a strong speckle noise. Thus, the proposed signal processing helps for the data interpretation, by improving the resolution and suppressing the noise. The observation model that relates the discontinuities to the data is modeled as a convolution process. The goal of the deconvolution processing is to recover the discontinuities from the data. So as to improve classi...