Pattern Matching on Vector Fields using Gabor Filter

J. Ebling and G. Scheuermann (Germany)


Vector Fields, Flow Visualization, Feature Detection, Ga bor Filter


Clifford convolution has been proposed as an operation for filtering vector fields with applications like pattern match ing. The development of a suitable Fourier transform al lows filtering in the frequency domain as well. Since spatial and frequency domain have both their advantages, an opti mally localized filter in both domains is of high interest. These filter are called Gabor filter in conventional image processing. Typical local pattern in flow fields like vortices and divergent flow can be described by a few coefficients in frequency space. Therefore, Gabor filter are well suited for the description of these patterns in the flow. We examine different approaches for pattern matching on vector fields using Gabor expansion and Gabor filter.

Important Links:

Go Back