No abstract

No abstract

In ultrasonic NDE of industrial components, the visibility of flaw echoes is corrupted by noise due to multiple scattering. Grain boundaries can reach the size of the same order of magnitude than discontinuities to be detected, becoming scatters of ultrasonic noise and could be confused with defect indications. Many studies have been conducted on the use of the wavelet theory for ultrasonic signal de-noising, but nothing has been done on the structural noise features and its analyzing wavelet function. In the framework of the automation of the ultrasonic signal analysis project, we have followed the exploration of the wavelet theory, from the continuous transforms to the discrete ones, and the experiments give us some ambivalent results. So for a best threshold control, our idea was directed to the investigation of the noise analyzing function. In this work the noise features were extracted by an energetic smoothing algorithm that allows the exploration of the noise analyzing function; by which the random nature of the noise in the spatial domain is bypassed. The energetic characterization of the noise and the defects allows an improved filtering process. The new smoothing algorithm performs an accurate signal reconstruction in an interesting computing time.

No abstract

No abstract