Wavelet transform-based approach for processing ultrasonic B-scan images

摘要

The attenuation of ultrasound beams in coarse-grained austenitic stainless steels due to grain boundary scattering influences flaw detection and sizing. This paper presents an approach for processing ultrasonic B-scan images of thick coarse-grained austenitic stainless steel parts using a discrete wavelet transform (DWT) technigue. In this approach, DWT processing of A-scan signals obtained during B-scan imaging is performed to reduce the noise caused by grain boundary scattering. The Daubechies (db8) mother wavelet with decomposition level 5 is found to be optimum for this purpose. Experimental studies are carried out on coarse-grained stainless steel specimens having flaws machined at different depths. These studies reveal that the proposed approach results in an improvement of more than 13 dB in signal-to-noise ratio, as compared to two-dimensional DWT processing of the original B-scan image using the same wavelet and decomposition level. Furthermore, the proposed approach produces a smaller number of artefacts than the two-dimensional DWT technique. The studies reveal that by using the proposed approach it is possible to detect flat-bottomed holes located at 40 mm below the surface of α 50 mm-thick stainless steel part with a grain size of 200 μm at normal incidence. The proposed approach is fast and attractive for online implementation.