Simulation of fractal multidimensional images using multidimensional recursive filters

Fractal multidimensional (MD) images may be generated by simulating MD fractal Brownian motion (fBm). This is usually achieved by appropriately weighting the magnitudes of the Fourier coefficients of the MD discrete Fourier transform (DFT) of the required fractal image. In this contribution, it is proposed that MD hyperspherically-symmetric recursive filters be used to approximate fractal images directly in the spatial domain, thereby allowing spatially-variant characteristics to be obtained without the undesirable edge effects of the MD DFT method. The method is used to generate realistic 2D fractal landscape images having spatially-variant characteristics.<
>