Complexity analysis of textures from urban images applying fractal dimension

In this paper, fractal dimensions based on the Bouligand-Minkowski and the Box-Counting 3D approaches were applied to estimate the complexity of textures from Google Earth’s images of São Carlos (SP), Brazil. Five pairs of urban images at different distances from the city center were selected. Figures 1 and 2 show the Bouligand-Minkowski and the Box-Counting 3D signatures, respectively. As we can see, the curves' behaviour of the latter can better distinguish the selected images than the former.

Fig.1: Bouligand-Minkowski Signature of the selected images


Fig.2: Box-Counting 3D signature of the selected images


The results show that as the distance to the city center increases, the higher the fractal dimension of the selected image is, and the more complex and chaotic the urban morphology is. This suggests that São Carlos presents a clear core-periphery structure.

For more details about these results see the short paper: Análise da complexidade de texturas em imagens urbanas utilizando dimensão fractal which was presented at IX Brasilian Symposium in Geoinformatics GEOINFO 2007 , in Campos do Jordão, Brazil.