Natural scene segmentation using fractal based autocorrelation

The authors propose a fractal based image segmentation model that is invariant to changes in incident light intensity. Since the fractals are inherently invariant under scale changes, this model therefore is a robust model for dynamic vision for mobile robot navigation in unstructured environments. The results of this model are presented as a classifier for natural textures. It is shown that the model meets the lower computation bound of surface based texture classification methods. The results of using this model for image segmentation are presented