Many objects properties are strongly determined by their geometric
properties: we only mention protein ternary structure. Application of
shape analysis range from model matching to clustering and model
segmentation, by investigating surfaces and characterizing important
features such as genus and boundaries. A common approach for shape
analysis is to reconstruct the surface; many of the algorithms are based
on Voronoi diagrams and Delaunay triangulations, since these define a
simplicial complex on sample data. Often, these algorithms invest
quadratic time in computing triangulations for sets as large (or even
larger) as the input data, and only in subsequent phases filter out the
results in a suitable manner; problems also arise when initial surface
has boundaries, holes or sharp edges. Our investigation tries to figure
out whether it is possible to reason about the genus of a surface
without relying on reconstruction, but rather reducing to a
combinatorial setting and avoiding geometric peculiarities.