Robust Techniques for Data Preprocessing
A summary on numerical cartography and digital photogrammetry, taking into account also some aspect of GIS’s and image processing, is presented, in order to list the principal problems of spatially referenced data analysis.
- PDF / 1,862,241 Bytes
- 18 Pages / 481.89 x 691.654 pts Page_size
- 27 Downloads / 237 Views
M.A. Brovelli, F. Migliaccio, L. Mussio and O. Sharif Polytechnic of Milan, Milan, Italy
ABSTRACT A summary on numerica! cartography and digital photogrammetry, taking into account also some aspect of GIS's and image processing, is presented, in order to list the principal problems of spatially referenced data analysis. As it is very well known, reliability and robustness are important properties, that may be assured in the data analysis by using suitable advanced techniques, but their introduction in the collocation method is at presant problematic. An attempt to introduce robust estimators in the collocation method, taking into account the different steps in which it is split, has been done and the results are here presented and discussed.
L. Mussio et al. (eds.), Data Acquisition and Analysis for Multimedia GIS © Springer-Verlag Wien 1996
60
M.A. Brovelli et al.
1. A VIEW ON GIS WORLD A GIS requires information from image processing, obtained by photogrammetric or remote sensors, as well as by means of hardcopy scanning of previously existing maps. Furthermore a GIS exchanges information with more general data bases. A GIS is often object oriented and these objects represent entities or entity classes. They get a spatial reference, while the most commonly used data banks do net get it. Attributes of various types are associated te the entities: they supply information about the entities both from the geometrica! and semantic point of view. The entity-attribute relations are characterized, if binary, by four cardinal values that represent the smallest and the largest numbers of entities in each direction of the relation. The geometrica! entities are regions, lines and points, while their attributes may represent different characteristics and they may constitute particular subentities. The georeference of each object allows for spatial (topological and geometrica!) analysis. In this way data integration and data fusion are possible and suitable mathematical models can be applied te study particular phenomena of different physical nature and semantic contents. The dimension of the data bases of a GIS requires a suitable organisation, with special regard te the structura of the data bases. A suitable methodology te implement the data bases and the set of entities of a GIS, that is used also in nonspatially referenced informative systems, is composed by four phases: • spatial; • conceptual; • logical; • physical. The first phase consists of a spatial object description; successively this description becomes abstract, se that the real world is described in a complete, non-ambiguous and of minimum redundancy way. This set of formal rules represents the conceptual scheme of the data, that is followed by the complex process of data modelling. Data exchange becomes quite immediate from the logical point of view, which gives some advantages, like the capability te supply information, relatively independent from hardware platforms and operative systems (software). Thus the physical implementation is only the last phas
Data Loading...
