Comparison of SAFT and two-dimensional deconvolution methods for the improvement of resolution in ultrasonic B-scan images

Abstract

This Paper compares the properties of the well known synthetic aperture focusing technique (SAFT) with those of two deconvolution methods, which have previously been less widely applied to digitized ultrasonic data. The linear deconvolution method of Wiener filtering and the non-linear maximum entropy method are applied to two-dimensional unrectified (RF) B-scan data, to improve the resolution simultaneously in the lateral and range directions. With SAFT, only an improvement in lateral resolution is possible. Data from linear scans of several unfocussed transducers over machined reflectors at different depths in steel were collected and used to quantify the resolution obtainable by the three processing methods. Wiener filtering is shown to give higher lateral resolution than SAFT processing, by factors of between 1.3 and 2, depending on the depth of the reflector. Maximum entropy gives even higher apparent lateral resolution, but has a tendency to underestimate reflector size. The improvement in range resolution obtainable with Wiener filtering was approximately a factor of 1.5, whereas maximum entropy gave corresponding factors of between 1.7 and 5. For far-field reflectors, deconvolution of rectified (envelope-detected) data is shown to give inherently lower lateral resolution than the deconvolution of the equivalent RF data.