cartographic images dozens of times and considerably decrease the reliability of
research by such maps.
The technical accuracy of research is connected to the reliability of measuring
instruments, calculation equipment, algorithms and programs, vulnerability of database
and methods of carrying out the measurements. In practice one tries to ensure that
measurement errors do not exceed the specified limits.
When determining the overall assessment of measurement results and calculations
by maps the cumulative impact of cartographic and technical accuracy is taken into
consideration.
In automation (computerization) cartographic generalization is distinguished by
two sides: semantic and geometric.
The generalization of qualitative characteristics takes place by combining
classification features of objects and combination of positions in the legend, taking into
consideration hierarchical structure of the phenomenon. It is done automatically using
single- or multiparameter classifications. The qualification selection of the objects is
the easiest realized in generalization.
The requirements of cartographic generalization impose strict limitations on
automation of geometric (spatial) component. Methods of formal (mechanical)
rounding of drawings are not suitable here, automatic recognition of the hierarchy of
geometric structures constitute the main task of computer image recognition.
Automation of generalization of data differs depending on the method of its
reflection – raster of vector. While generalizing raster data their values are recalculated
taking into consideration adjacent pixels. Algorithms operate simultaneously with
qualitative and quantitative characteristics of statistical or attributive data.
In the generalization of the lines geometric manipulation with chains of
coordinate pairs (х, у) occurs. Computer operations coincide by content with the
traditional procedures of cartographic generalization.
Simplification — means operations which allow to exclude extra coordinate pairs
based on the specified geometric criterion (e.g. distance between points or offset from
the central line).
Rounding — provides a shift of coordinate pairs to eliminate small gyrus and
store only the biggest bends of the lines. As a rule, rounding of digitalized line is used
to decrease distortions of coding, meaning that it considerably improves such a line.
Object shift — performs the shift of objects so that while zooming out, to avoid
their floating or overlapping; most of algorithms of object shifts in vector format is
oriented on interactive mode.
Correction — some details are introduced to a simplified set of data (e.g., to
emphasize the peculiarities of the coast); one of the methods proposes dividing the line
by introduction of points to achieve similarity to the original.
- 277 -