Fractal Topological Analysis for 2D Binary Digital Images

Abstract

Fractal dimension is a powerful tool employed as a measurement of geometric aspects. In this work we propose a method of topological fractal analysis for 2D binary digital images by using a graph-based topological model of them, called Homological Spanning Forest (HSF, for short). Defined at interpixel level, this set of two trees allows to topologically describe the (black and white) connected component distribution within the image with regards to the relationship "to be surrounded by". This distribution is condensed into a rooted tree, such that its nodes are connected components determined by some special sub-trees of the previous HSF and the levels of the tree specify the degree of nesting of each connected component. We ask for topological auto-similarity by comparing this topological description of the whole image with a regular rooted tree pattern. Such an analysis can be used to directly quantify some characteristics of biomedical images (e.g. cells samples or clinical images) that are not so noticeable when using geometrical approaches.

Structural Fractal Analysis of the Active Site of Acetylcholinesterase in Complexes with Huperzine A, Galantamine, and Donepezil

“…The key quantitative characteristic of fractal objects is fractal dimension D. Therefore, a significant number of modern works are devoted to the calculation and analysis of this quantity in various fields of knowledge, including physics, [22,23] geophysics, [24,25] chemistry, [26,27] biology, [28] medicine, [29,30] economics, [31,32] materials science, [33] and computer science. [34,35] Another important characteristic of fractals is Hurst exponent H, which is used as the measure of a long-range dependence in time and spatial series and changes from 0 to 1. [36] In, [37] we proposed to consider a molecular interatomic-distance histogram (IDH) as a fractal object.…”