Interactive Image Processing and Pattern Recognition of Digitized Flow Patterns

Abstract

In this thesis report, interactive algorithms aid in the analysis of fluid flows are presented. Special functions and algorithms to average, smooth, and calculate the similarity between digitized curves were developed. The developed routines process the images in the spatial domain, thereby eliminating the need to calculate discrete Fourier and inverse Fourier transforms. The algorithms are effective, efficient, and fast. As an integral part of the algorithms, special data buffer routines for the effective data manipulation of curves, as well as cursor routines, were developed. An arbitrary set of frames consisting of curves, or an arbitrary set of curves, can be averaged or smoothed. Curves can be smoothed with a modified, variable, convolution filter. A special function makes it possible to express the similarity of two curves in a numerical value. This technique can be used to study time effects in fluid flows. Noise reduction can be obtained by averaging and smoothing a set of curves.