Aim of this thesis was to present an overview of fractal geometry and its perspectives in geoscience, espetially in settlement studies
Since Mandelbrot (1967) published its basics, fractal geometry and fractal dimension
(non-integer dimension) is well known as a valuable tool for describing the shape of
objects. It gained large popularity in many fields of natural sciences, including e.g.
ecology, geography, GIScience, where the measures of object’s shape are essential.
One of the major principles in fractal geometry is self-similarity and self-affinity. The
most theoretical fractal objects, such as Barnsley’s fern, are self-similar (any part of the
object is exactly similar to the whole) and self-affine (transformed self-similar objects).
In this thesis, three case studies ware performed
The process of urbanisation of the city of Olomouc was examined by the means of fractal geometry.
An interesting connection between build-up area and its fractal dimension was found and described.
Also fractal dimension of the edge of the city was measured, however the results were not conclusive.
Second case study dealt with possibilities of automated classification based on fractal characteristics.
Leaves of seven species of three families were examined and classified.
For leaves, we show that discrimination based on only two fractal features has more
than 90% accuracy. This notion is important, because it proves that automated
classification can be based also on complexity of shapes and not only on their qualitative
measurements. Fischer discriminant analysis is used to distinguish between families and
species with satisfactory results.
At last, the results from the second case study were applied to city landuse of the Olomouc city
Unfortunately, the classification was unsuccesful, which shows, that different types of city land use can not be distinguished by their geometric characteristics.