Detection of vortices in flow images using the wavelet transform

In this paper a novel wavelet based scheme for detecting vortices in flow images is proposed. This work considers the local salient structures, vortices, are blurred by the complex flow structures, or by the diffusion of dye or smoke in experimental flow visualization images. The wavelet transform acts as a feature discriminator to separate vortices from the blur or other complex flow structures. In the decomposed flow images computed by wavelet analysis, vortices corresponding to the proper scale are enhanced while the uninteresting background is suppressed. With the help of computer vision techniques, the enhanced vortices are localized. Experimental results show that the proposed approach outperforms a vector field estimation approach.