Applications of transform techniques in ultrasound imaging and profilometry

Abstract

This thesis consists of two parts. The first part describes the studies in the ultrasound pulse-echo frequency domain imaging and the second part presents the studies at the non-contact measurement of 3-D object surface using Fourier transform profilometry. In the first part, the ultrasound backscattering characteristics of some simple phantoms were examined by two methods. These were ‘pure’ frequency domain ultrasound imaging based on peak frequencies in a series of echo spectra (or spectrograms) and the new type B mode frequency domain imaging based on both time domain and frequency domain information. The two major techniques used to analyse the time-varying echo signals, the short-time Fourier transform (STFT) and the orthogonal-like Gabor transform (OLGT), were studied and compared in details in order to generate more accurate spectrograms for imaging. The experimental results have found that the frequency domain images have better lateral resolution than the conventional B mode images. By using the appropriate short-time sliding window, the axial resolution in the frequency domain images can be improved. The new type B mode frequency domain images, which not only show the physical sizes of the scanned objects but also contain the frequency information of scattered targets, may have more potential diagnostic value in the future. The second part describes how to improve the accuracy of non-contact measurement of three-dimensional (3-D) object surface by using the new image processing-based methods in Fourier transform profilometry. The methods include application of the weighted window function to the isolated first sidelobe in frequency domain, automatic determination of the bandwidth of bandpass filter based on the actual spectrum and use of object boundary guided phase unwrapping to solve large-scale discontinuities around the object boundary on the modulo 2π phase map. The accuracy of measurement of the 3-D object surfaces has been improved using those new image processing-based methods.