No label defined (Q4337020)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1010538
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 1010538 |
Statements
19 May 1997
0 references
morphological skeleton
0 references
geographic information systems
0 references
0.8037942051887512
0 references
0.7617558836936951
0 references
0.7444414496421814
0 references
This work is concerned with the decision problem of determining topological relations among two-dimensional simply connected regular closed compacta in \(\mathbb{R}^2\) from uncertain data (measurements) and therefore falls into the province of approximate reasoning. Topological relations among objects are represented by means of \textit{Egenhofer's} model [cf. Geomatica 47, 261-273 (1993)] in which one considers for two objects in question, A, B, eight basic relations: disjunct, touch, overlap, covers, covered by, contains, contained by, equal; they are represented by means of binary intersection matrices whose entries are characteristic function values of intersections of any pair of sets from the product: \{interior A, boundary A, complement A\} \(\times\) \{interior B, boundary B, complement B\}. In the set of relations, one introduces a natural partial order based on topological transformations among relations; intersection matrix representation allows for a metric in the space of relations based on XOR applied entrywise. In the set of relations one also distinguishes dominant ones on the basis of their behaviour during topological transformations: they hold only momentarily in time. As the result, the graph of relations is split into two subgraphs (clusters): \{disjunct, touch, weak overlap\}) with the dominant touch and \{equal, covers, covered by, contains, contained by, strong overlap\} with the dominant equal. The author proposes statistical reasoning (maximum-likelihood classification) as the tool to determine the relation which holds for two given objects perceived through measurement, selecting a dominant relation as the null hypothesis versus a dominated one as an alternative. To this end, he introduces the morphological distance between objects based on the notion of a morphological skeleton, and he obtains a relationship among classes of this distance and clusters of relations (Tab. 4.1). The maximum likelihood classification is applied to select on the basis of the evidence \(e\) the most likely distance class \(\omega^\wedge\) from which the decision about the relation is read. To estimate the necessary probabilities \(P(\omega|e)\), Bayesian reasoning is applied. The work is concluded with a section (Sect. 6) in which examples of applications are discussed and with a brief discussion (Sect. 7) of open problems and limitations of the method proposed. Appendices (A--D) enclosed in the end, provide the reader with basic notions from topology, mathematical morphology, metric theory and convolution theory relevant to the content of the work. This work is an interesting example of a process of approximate reasoning which combines morphological and statistical tools into a hybrid system capable of decision-making in the area of geographic information systems.
0 references