feasibility study for an automated generalisation ...€¦ · feasibility study for an automated...
TRANSCRIPT
Feasibility study for an automated
generalisation production line for
multiscale topographic products
Ron Nijhuis1, Marc Post1, Vincent van Altena1, Jan Bulder1,
Ben Bruns1, John van Smaalen4, Jantien Stoter1,2,3
1Kadaster, Apeldoorn
2GISt, OTB, TU Delft
2Geonovum, Amersfoort
4Esri NL, Rotterdam
The Netherlands
Content
• Why?
• Methodology
• Findings
• Work in progress
Background
• Necessity to automate production line of Kadaster
• Assumption that today’s user might be more flexible
with the product
• See what is possible for a technical point of view and
refine the product in consultation with the user
• This presentation focuses on technical feasibility
Why?
Methodology
Findings
Work in progress
Scope
• Focus on:
– fast retrieval small scale version of TOP10NL
updates
– cartographic representations
• Focus on static visualisation on screen or paper
• Fully automated
• Start with enriched data set
• Pilot: focus on 1:50k scale
Why?
Methodology
Findings
Work in progress
Methodology
• Extend research: Stoter et al, 2009, Specifying map
requirements for automated generalisation of topographic data,
The Cartographic Journal Vol. 46 No. 3 pp. 214–227
– Implement guidelines for interactive
generalisation in trial and error
– Compare results with existing maps
– Enrich source data and guidelines
• ArcGIS 10 environment
Why?
Methodology
Findings
Work in progress
Automated generalisation workflow
For producing 1:50k map from (enriched)
TOP10NL:
1. Data generalisation
2. Symbolisation
3. Cartographic generalisation
Why?
Methodology
Findings
Work in progress
1. Data generalisation (1/3)
• TOP10NL codes are recoded into TOP50 codes
• Urban areas detected
• Areas with many buildings converted to built-up
– Symbols for important buildings are kept
• TOP10NL polygons (centrelines) converted into
TOP50 centrelines: to be used for smaller scales!
Why?
Methodology
Findings
Work in progress
1. Data generalisation (2/3)
• Road network is pruned with a thinning road
algorithm; before:
– Deletion of:
• cycle paths parallel to roads
• non-paved dead-ends shorter than 1 km
in rural areas
– Selection of:
• free cycle paths
• access roads to buildings
Why?
Methodology
Findings
Work in progress
1. Data generalisation (3/3)
• Land use parcels extended to new centrelines
• Water network pruned by:
– removing small water parallel to roads
– selecting remaining free lying water
– drainage information (added)
• Too small water and land use polygons deleted if
isolated otherwise amalgamated
• Main railway network detected
Why?
Methodology
Findings
Work in progress
Spoor Top10NL
3. Cartographic generalisation
• Symbolised dikes, railways, roads, water, land
use (linear objects & area boundaries) displaced
• Land use polygons recreated from displaced
linear object and former terrain codes assigned
• Remaining buildings simplified and displaced to
avoid overlap and to meet minimum size
Why?
Methodology
Findings
Work in progress
Buildings in TOP10NL
Buildings in TOP10NL
simplified
Buildings discplaced
Source data: TOP10NL
1:50k target map, obtained fully
automatically
Results (1/2)
Why?
Methodology
Findings
Work in progress
1:50k target map, obtained fully
automatically
1:50k target map obtained interactively
Results (2/2)
Work in progress (1/2)
• Errors in the source data, e.g. roads
• Pruning of water:
– Network needs to be built: for smaller scales
• Generalisation of additional objects
• Annotation
• Study if and which information should be added at
smaller scales
• Partitioning
Why?
Methodology
Findings
Work in progress
Work in progress (2/2)
• Further evaluate and fine tune 1:50k pilot results
(still too much detail)
• Discuss with users intermediate results and
consequences (e.g. repeatability; updates)
• Extend to other test areas to validate
• Extend to other scales (which ones?)
• Study star versus ladder approach
• Change the law
Why?
Methodology
Findings
Work in progress
Future
• Replace current production line
• Change the law