Fig 3.

Wavelet analysis method demonstrated for a Brownian motion. The signal f(t′) (a) is convolved with Mexican hat filters g(t′) (b) to yield the wavelet transforms W(t,a) (c). The filters were generated with scale parameters, a, from bottom to top, of 1, 2, 4, 8, and 16. There is a power law relationship between the variance of the transforms, V, and effective filter width, λ, indicating that the signal is self-affine with slope, β, equal to 1.9 (d).