4 Progressive Techniques
79
Cecconi and Weibel [10] describe a framework for map generalization to be
applied for on-demand Web mapping. Rather than structure spatial data with respect
to scale-dependent transmission and display, they suggest adapting existing generalization algorithms for on-the-fly generalization and develop new ones to provide for
missing generalization functionality.
In [11] an adaptive zooming method for Web mapping is presented. The database
is organized as a number of LOD and groups of selected feature classes. The LOD are
derived from pre-selected scales, while the groups of selected feature classes are generalized on-the-fly. Given a user zooming request with a specific target scale, some
objects at the corresponding LOD are retrieved from the database. The remaining objects are dynamically created by on-the-fly generalization. Different generalization
operations (including those involving changes in the map topology) are considered.
This results in vector maps with proven cartographic quality. Since on-the-fly generation applies only to a small number of objects instead of the entire map, this approach
is applicable in real time. However, it requires the a priori selection of which features
are to be stored in the database and which are to be generalized on-the-fly.
Finally, recently a few Web mapping systems have been developed using objectoriented databases and multi-agent technology. One such system, called LAMPS2
[26] relies on an object-oriented multiscale database and applies multi-agent technology to support automated map generalization (including operations that change
the map topology). Although the system shows to be promising, in its current version
it does not represent a solution to real-time online generalization. In addition, neither
a progressive data delivery approach nor the problem of avoiding redundancy caused
by multiple geometry objects storage have been discussed in relation to the LAMPS2
system.
4.6 Research and Implementation Issues
As seen in previous sections, research on progressive vector map transmission mostly
relates to map generalization techniques that allow for coarser representations to be
generated, each one at different LOD.
Although map generalization and multiple map representations have been studied
for several years by the GIS community, several research and implementation issues
remain unsolved for the successful online application of progressive vector map data
transmission.
More specifically, based on our experience in the development of progressive
transmission systems and on a thorough analysis of the literature, we have identified
some main research priorities that require much investigation. These are summarized
below:
• Real-time Generalization and Data Compression: A complete formalization
of map generalization principles and practice is still lacking; more efforts are
required to develop new algorithms for support to automated generalization.
79
Cecconi and Weibel [10] describe a framework for map generalization to be
applied for on-demand Web mapping. Rather than structure spatial data with respect
to scale-dependent transmission and display, they suggest adapting existing generalization algorithms for on-the-fly generalization and develop new ones to provide for
missing generalization functionality.
In [11] an adaptive zooming method for Web mapping is presented. The database
is organized as a number of LOD and groups of selected feature classes. The LOD are
derived from pre-selected scales, while the groups of selected feature classes are generalized on-the-fly. Given a user zooming request with a specific target scale, some
objects at the corresponding LOD are retrieved from the database. The remaining objects are dynamically created by on-the-fly generalization. Different generalization
operations (including those involving changes in the map topology) are considered.
This results in vector maps with proven cartographic quality. Since on-the-fly generation applies only to a small number of objects instead of the entire map, this approach
is applicable in real time. However, it requires the a priori selection of which features
are to be stored in the database and which are to be generalized on-the-fly.
Finally, recently a few Web mapping systems have been developed using objectoriented databases and multi-agent technology. One such system, called LAMPS2
[26] relies on an object-oriented multiscale database and applies multi-agent technology to support automated map generalization (including operations that change
the map topology). Although the system shows to be promising, in its current version
it does not represent a solution to real-time online generalization. In addition, neither
a progressive data delivery approach nor the problem of avoiding redundancy caused
by multiple geometry objects storage have been discussed in relation to the LAMPS2
system.
4.6 Research and Implementation Issues
As seen in previous sections, research on progressive vector map transmission mostly
relates to map generalization techniques that allow for coarser representations to be
generated, each one at different LOD.
Although map generalization and multiple map representations have been studied
for several years by the GIS community, several research and implementation issues
remain unsolved for the successful online application of progressive vector map data
transmission.
More specifically, based on our experience in the development of progressive
transmission systems and on a thorough analysis of the literature, we have identified
some main research priorities that require much investigation. These are summarized
below:
• Real-time Generalization and Data Compression: A complete formalization
of map generalization principles and practice is still lacking; more efforts are
required to develop new algorithms for support to automated generalization.