It is a widely observed phenomenon in computer graphics that the size of the silhouette of a polyhedron is much smaller than
the size of the whole polyhedron. This paper provides, for the first time, theoretical evidence supporting this for a large
class of objects, namely for polyhedra or, more generally, tessellated surfaces that approximate surfaces in some reasonable
way. The approximated surfaces are two-manifolds that may be nonconvex and nondifferentiable and may have boundaries. The
tessellated surfaces should, roughly speaking, have no short edges, have fat faces, and the distance between the mesh and
the surface it approximates should never be too large. We prove that such tessellated surfaces of complexity n have silhouettes of expected size
where the average is taken over all points of view. The viewpoints can be chosen at random at infinity or at random in a bounded
region. 相似文献
To detect database records containing approximate and exact duplicates because of data entry error or differences in the detailed
schemas of records from multiple databases or for some other reasons is an important line of research. Yet no comprehensive
comparative study has been performed to evaluate the effectiveness of Silhouette width, Calinski & Harbasz index (pseudo F-statistics)
and Baker & Hubert index (γ index) algorithms for exact and approximate duplicates. In this paper, a comparative study and effectiveness of these three
cluster validation techniques which involve measuring the stability of a partition in a data set in the presence of noise,
in particular, approximate and exact duplicates are presented. Silhouette width, Calinski & Harbasz index and Baker & Hubert
index are calculated before and after inserting the exact and approximate duplicates (deliberately) in the data set. Comprehensive
experiments on glass, wine, iris and ruspini database confirms that the Baker & Hubert index is not stable in the presence
of approximate duplicates. Moreover, Silhouette width, Calinski and Harbasz index and Baker & Hubert indice do not exceed
the original data indice in the presence of approximate duplicates. 相似文献
An approach for the characterisation of the groundwater system of the southern plain of Friuli-Venezia Giulia Region (Italy) is proposed on the basis of its physico-chemical composition, in order to detect multivariate patterns for unpolluted waters typical of specific areas in the plain, as well as for eventual polluted zones. The analytical data are relative to 38 wells (depth ranging from 20 to 200 m) sampled in three different periods along a year. Ten physico-chemical parameters were determined: conductivity, temperature, dissolved oxygen, calcium, magnesium, chlorides, nitrates, sulphates, atrazine and desethylatrazine.
Cluster analysis (CA) provides the methodological bases for detecting the classes of freshwater being typical for the considered plain: partitioning around medoids (PAM) and fuzzy clustering are considered. The number of classes to be characterised and the clustering algorithm are selected by comparing the average silhouette index for models counting from 2 to 10 clusters; six classes obtained by PAM partition the data set at best.
Plotting the frequencies of cluster membership for each well on a map permits the association of the six classes of waters to five easily recognisable geographical areas and to one group of two wells that are highly polluted by nitrates and triazines.
Averages and ranges of values for physico-chemical parameters of each class can be provided according to this methodology, defining a set of values being characteristic for the composition of waters belonging to the classes of wells identified in the considered plain. 相似文献