Raster digitization is performed by means of a solid state linear array consisting
of 1024 elements (element size 12.5 x 12.5 micrometers). The scanning of the
document is performed in strips, where the map format and the raster pixel
resolution determine the number of strips.
The raster data is recorded in a raster, two-dimensional-data-structure on a
magnetic tape. Since raster scanning generates large volumes of data, the "run-
length encoding" data compression scheme is used. Run length encoding stores a
counter for every continuous string of equal values, often indicated by 0 or 1 bit
indicators. This avoids having to store every data location.
The stored raw raster data may be reviewed on the graphic display terminal to
ensure that the entire image have been captured and that the quality standards
are met.
For the generation of the initial data base and to be able to exploit the data by
traditional digital mapping systems, the raster data has to be converted to a
vector format.
The raster-to-vector conversion follows three steps:
- skeletonization (or line thinning): which reduces the raster elements to
one unit of resolution, located in the geometric center of the
cartographic feature,
- line-extraction: the identification of unique line segments, and
- topology reconstruction: which refers to the explicit identification of the
spatial relations of points, lines and areas in the vector file.
Also raster-to-vector conversion software can differentiate between linear
features and cartographic areal features. In the case of solid areas the
vectorizing software detects the edges.
The raw vector data file is reviewed on the graphic display workstation for data
quality completeness, and the detection and correction of data anomalies. This
process can be accomplished interactively via computer graphics techniques or by
means of post-processing software. The software is used for the detection and
editing of connected or unconnected spikes, correction of "nodes" resulting from
the intersection of two or more lines, removing bridges and unwanted connection,
etc.
Upon completion of the post processing, the data files are not error free, and the
features are not tagged with their proper cartographic codes and attributes. To
perform the editing task, the files are transferred to the editing graphic
workstation. The amount of editing required depends on the complexity of the
graphic, the quality of the line work and the generic type of the cartographic data
(e.g. non-intersecting lines, merging or crossing networks, etc).
For contour overlays (non-intersecting features), the editing is minimal and
consists mainly of detecting coalescence (bridges) between adjacent contour lines,
closing long gaps due to contour labelling, etc.
Overlays where features are merging and intersecting, e.g. hydrography or
transportation, pose additional problems because of the multi-nodes created at
the intersections of lines with varying thickness.
517