Skip to Main Content
A pair of divergent air-coupled capacitive ultrasonic transducers (CUTs) with curved backplates was used to acquire signals through regions of air containing solid objects, air flow, and temperature fields. Fan-beam datasets were collected and used in a tomographic reconstruction algorithm to produce cross-sectional images of the area under interrogation. In the case of the solid objects, occluded rays from the projections were accounted for using a compensation algorithm and a priori knowledge of the object. A rebinning routine was used to pick out parallel ray sets from the fan-beam data. The effects of further reducing the number of datasets also were investigated, and, in the case of imaging solid objects, characteristic Gibbs phenomena were seen in the reconstructions as expected. However, when imaging temperature and flow fields, the aliasing artefacts were not seen, but the reconstructed values decreased with the size of dataset used. The effect of changing the kernel filter function also was investigated, with the different filters giving the best compromise between image noise, reconstruction accuracy, and amount of data required in each scenario.