Segmentation of textures with different roughness using the model of isotropic two-dimensional fractional Brownian motion

An attempt is made to apply a novel multidirectional resolution analysis to obtain boundaries between areas with different grades of 'roughness'. To this end, the image texture is parametrized with respect to direction, based on the model of isotropic 2-D fractional Brownian motion (2-D FBM). Some theoretical aspects of this model have not been clarified yet. The authors first provide some of the crucial missing links in the theory of 2-D FBM and present a concrete realization of this stochastic process. They then describe the segmentation procedure in detail. First, they present a suboptimal but fast estimation algorithm to obtain angle-dependent fractal parameter subimages, and compare its performance with the Cramer-Rao bound. They then apply matched edge-detection procedures to each subimage separately, and combine the angle-dependent boundary pixels to obtain the texture boundaries of the entire scene.<<ETX>>