5 Model Generalization
95
exact new tolerance value. The worst-case estimation of the new top tolerance value
‘err ij,’ according to the formula given in [20] page 92 and reformulated in Fig. 5.9
as ‘err ij = dist(point(ij), line(b i, e j) + max(err i, err j)’ only uses the top-level
information of the two participating trees.
An improvement, which keeps the structure of the merged BLG-tree unaffected
(so the lower level BLG-tree can be reused again), is to compute the exact tolerance
value (‘err ij exact’) of the new approximated line, which is less than or equal to the
estimated worst-case (‘err ij’). In Fig. 5.9, this would be the distance from point 5
of edge ‘j’ to the dashed line. This tolerance value would be ‘1.1,’ which is less than
the worst-case estimate of ‘1.4’ for the tolerance. The drawback is that one has to
descend in the lower level BLG-tree for the computation (this may be a recursion)
and this will take more time (but normally has to be done only once: during the
generalization process when the tGAP structure is built). The advantage is that during
the use of the structure (which happens more often than creation) one has a better
estimate and will not descend the BLG-tree unneeded. For example, assume one
needs a tolerance of ‘1.2,’ then with the worst-case estimate one has to descend into
the two child BLG-trees. This is not necessary when the top-level tolerance value for
the joined BLG-tree is computed instead of estimated.
When the data set is accessed and the faces and edges are selected based on
their importance range, the corresponding BLG-trees are used for line simplification.
Depending on the requested tolerance value the (joined) BLG-tree is traversed in
order to produce the appropriate detail level. Note that the relationship between the
tolerance value and map scale is quite direct (e.g. one could use the size of a pixel on
the display screen as the tolerance value).
5.4 Current Internet GIS Protocols
One of the main features of the proposed structure is that it supports progressive
transfer and smooth zooming in a Web service/client context. Therefore in this section we will look at the most relevant Web service Protocols for vector data. We also
point out the main characteristics of a Web service/client architecture.
5.4.1 Geo-Web Services
For Web services that provide access to geo-information the standardization efforts
of the open geospatial consortium (OGC) are very important, therefore some background is presented here first.
The OGC was founded in 1994 by a number of software companies, large
data providers, database vendors, and research institutions. It can be considered an
industry-wide discussion and standardization forum for the geo-application domain.
One of the goals of OGC is to enhance interoperability between software of different vendors. With this purpose also a number of Web service interface specifications
have been developed, the first two were the Web map service (WMS) and the Web
feature service (WFS) protocols.
95
exact new tolerance value. The worst-case estimation of the new top tolerance value
‘err ij,’ according to the formula given in [20] page 92 and reformulated in Fig. 5.9
as ‘err ij = dist(point(ij), line(b i, e j) + max(err i, err j)’ only uses the top-level
information of the two participating trees.
An improvement, which keeps the structure of the merged BLG-tree unaffected
(so the lower level BLG-tree can be reused again), is to compute the exact tolerance
value (‘err ij exact’) of the new approximated line, which is less than or equal to the
estimated worst-case (‘err ij’). In Fig. 5.9, this would be the distance from point 5
of edge ‘j’ to the dashed line. This tolerance value would be ‘1.1,’ which is less than
the worst-case estimate of ‘1.4’ for the tolerance. The drawback is that one has to
descend in the lower level BLG-tree for the computation (this may be a recursion)
and this will take more time (but normally has to be done only once: during the
generalization process when the tGAP structure is built). The advantage is that during
the use of the structure (which happens more often than creation) one has a better
estimate and will not descend the BLG-tree unneeded. For example, assume one
needs a tolerance of ‘1.2,’ then with the worst-case estimate one has to descend into
the two child BLG-trees. This is not necessary when the top-level tolerance value for
the joined BLG-tree is computed instead of estimated.
When the data set is accessed and the faces and edges are selected based on
their importance range, the corresponding BLG-trees are used for line simplification.
Depending on the requested tolerance value the (joined) BLG-tree is traversed in
order to produce the appropriate detail level. Note that the relationship between the
tolerance value and map scale is quite direct (e.g. one could use the size of a pixel on
the display screen as the tolerance value).
5.4 Current Internet GIS Protocols
One of the main features of the proposed structure is that it supports progressive
transfer and smooth zooming in a Web service/client context. Therefore in this section we will look at the most relevant Web service Protocols for vector data. We also
point out the main characteristics of a Web service/client architecture.
5.4.1 Geo-Web Services
For Web services that provide access to geo-information the standardization efforts
of the open geospatial consortium (OGC) are very important, therefore some background is presented here first.
The OGC was founded in 1994 by a number of software companies, large
data providers, database vendors, and research institutions. It can be considered an
industry-wide discussion and standardization forum for the geo-application domain.
One of the goals of OGC is to enhance interoperability between software of different vendors. With this purpose also a number of Web service interface specifications
have been developed, the first two were the Web map service (WMS) and the Web
feature service (WFS) protocols.