Image processing techniques for Doppler global velocimetry
There is a demand for a whole field velocimetry technique which offers the capability of rapid characterisation of complex engineering flow fields. This thesis describes a research programme aimed at the development of a reliable Doppler global velocimeter, suited to the measurement of such flows. The programme of work undertaken is discussed with reference both to research undertaken elsewhere and to previous work on the system at Oxford. While much of the underlying technology required for the construction of an accurate and reliable velocimeter has already been studied in Oxford and elsewhere, little attention has been paid by previous workers to the examination of the impact of data processing techniques on attainable flow measurement accuracy. In the present work, a number of image processing methods have been utilised for Doppler global velocimetry data processing. Those methods are described here, together with a theoretical analysis of their expected performance when applied to Doppler global velocimetry data. The expected error resulting from image processing considerations and also from the physical characteristics of the Doppler global velocimetry hardware are quantified in such a way that error estimates may be computed for real measured data frames. The results of the application of the velocimeter to the simple test case of measuring a velocity component of a rotating disc are presented. The velocimeter was subsequently applied to the measurement of a free jet flow and to a transonic flow field in a convergent-divergent nozzle. Correlations with accepted velocity field values were undertaken, and compared to the expected error previously determined. The choice of image processing algorithms was found to be of great importance in terms of Doppler global velocimetry measurement accuracy.