principles of 5d modeling - geospatial world forum van oosterom.pdfmay 05, 2012 · principles of...
TRANSCRIPT
5-5-2012
Challenge the future
Delft University of Technology
Principles of 5D modeling
Peter van Oosterom,
GIS technology, OTB, Delft University of Technology
Seminar: 5D Modeling, 27 April 2012, GWF Amsterdam
2 Principles of 5D modeling
Contents
• Introduction
• 3D space+time example: 4D Cadastre
• Scale as dimension
• Conclusion
3 Principles of 5D modeling
Previous research: separate
treatment of 3D, Scale and Time
Time
(Kraak, 2006) Scale
(Vermeij, 2003)
3D (Penninga, 2007)
4 Principles of 5D modeling
Approach: 3 iterations
5 Principles of 5D modeling
• Multidimensional polyhedra
• GISt, TU Delft: Arens, Stoter (2004)
• Regular Polytopes
• GISt, TU Delft: Thompson (2007)
• Simplical Homology
• GISt, TU Delft: Penninga (2008)
Method: Apply mathematical theories on
multi-dimensional data modelling
6 Principles of 5D modeling
Poincaré simplicial homology
Solid mathematical foundation:
A n -simplex Sn is defined as smallest convex set in
Euclidian space Rm of n+1 points v0 , …, vn
(which do not lie in a hyper plane of dimension less than n)
7 Principles of 5D modeling
Poincaré simplicial homology
The boundary of simplex Sn is defined as sum of (n-1) dimensional
simplexes (note that ‘hat’ means skip the node):
Sn =
remark: sum has n+1 terms
ni
n
i
i vvv ,...,ˆ,...,)1( 0
0
8 Principles of 5D modeling
Contents
• Introduction
• 3D space+time example: 4D Cadastre
• Scale as dimension
• Conclusion
9 Principles of 5D modeling
Integration of 3D+time: 4D Cadastre
• In addition to spatial (3D) aspect, rights, restrictions and responsibilities include a temporal aspect
• To be able to manage the dynamics in land administration the
time (fourth) dimension must be handled as well
10 Principles of 5D modeling
Conceptual Cadastre Basis
2D: a planar partition of the surface
3D: a partition of space with no overlaps or gaps
4D: no overlaps or gaps in the rights, not only in space but also in parallel the time dimension
Partition: no gaps or overlaps in the parcelation on which the rights are based
11 Principles of 5D modeling
Implementation
X
Y
t
t0
t1
t2
State 1 t0 to t1 – 1 parcel
State 2 t1 to t2 – 3 parcels
State 3 t2 to now – 4 parcels
now
2D: partition of the surface based on a 2D topology with
faces, edges and nodes
3D: partition of space based on a complete 3D topological structure based on volumes, faces, edges and nodes
4D: use a 4D space-time topological structure
12 Principles of 5D modeling
3D Tunnel registration in Queensland
13 Principles of 5D modeling
River is moving over time and legal Boundary follows (true 4D)
14 Principles of 5D modeling
More cases:
Timesharing
• 3D volumetric
survey plan
(apartments)
• Timesharing of
40 units/week:
40*52 shares
• Timeshare can
be traded,
mortgaged, etc.
• 3D+time=4D
15 Principles of 5D modeling
16 Principles of 5D modeling
4D cadastre: separate space and time
or an integrated attribute?
• Advantages of separate attributes: 1. Already able to represent all cases
2. Supported by state-of-the art technology
3. Temporal aspect is more than just one dimension
• Advantages of integrated 4D data type: 1. optimal efficient 4D searching
2. Parent-child becomes topology neighbor query in time
17 Principles of 5D modeling
P1
P2 P3
P5
P4
t2
t1
t0
time
y
x
Subdivision
of parcels
18 Principles of 5D modeling
4D data type advantages (cont.)
• Advantages of integrated 4D data type: 1. optimal efficient 4D searching
2. Parent-child becomes topology neighbor query in time
3. Foundation of full (4D) partition: no overlaps or gaps in
space and/or time
4. 4D analysis: do two moving cattle rights have spatio-
temporal overlap/touch
19 Principles of 5D modeling
t2
t1
t0
time
y
x
P2 P1
Moving
cattle
20 Principles of 5D modeling
Contents
• Introduction
• 3D space+time example: 4D Cadastre
• Scale as dimension
• Conclusion
21 Principles of 5D modeling
1:10.000 1:100.000 1:50.000
1:250.000 1:500.000
Context of the research
22 Principles of 5D modeling
Early use of additional dimension for
scale (importance) representation
• Alternative Reactive-tree
(van Oosterom,
Auto-Carto 10, 1991)
23 Principles of 5D modeling
Generalized Area Partitioning-tree
24 Principles of 5D modeling
Generalized Area Partitioning-tree
(GAP-tree) history
• Normal GAP-tree (van Oosterom 1993) areas are stored as independent polygons computed redundancy (both at given scales and between scales)
• Vermeij et al. 2003 proposed topological GAP-tree: edges and faces (with importance range, consider as height), reduced redundancy between neighbors scale/imp with 3D prisms
25 Principles of 5D modeling
Contents
• Introduction
• 3D space+time example: 4D Cadastre
• Scale as dimension
• Conclusion
26 Principles of 5D modeling
Conclusions
• Vario-scale nD maps based on (n+1)D representations and
slicing (selecting) with hyperplanes:
• tGAP structure translates 2D space and 1D scale in an
integrated 3D topological representation: no overlaps and no
gaps (in space and scale)
• Starting with 3D space and adding scale results in 4D
• Starting with 3D space and time (history) and adding scale
results in 5D topological structure (again no gaps/overlaps in
space, time or scale), well defined neighbors in space, time
and scale directions
27 Principles of 5D modeling
Questions?
28 Principles of 5D modeling
patent pending nr. OCNL 2006630
prepared by Dirk de Jong, European Patent Attorney, Vereenigde
ICA 14th Generalisation Workshop, 30 June-1 July 2011, Paris, France
Towards a true vario-scale structure
supporting smooth-zoom
29 Principles of 5D modeling
Non-flat slice mixed scale map
(fish-eye example)
x
y
S=0
S=0.5
source: Harrie et al, 2002, ISPRS Archives 34(4):237–242
30 Principles of 5D modeling
Non-horizontal slice mixed scale map