Efficient Navigation in Temporal, Multi-Dimensional Point Sets (April 2013)
1. 1Challenge the future
Efficient Navigation in Temporal,
Multi-Dimensional Point Sets
Efficiёnte navigatie in tijd-afhankelijke, multi-dimensionale
puntengegevens
Christian Kehl
2. 2Challenge the future
General Introduction
• Plenty of point set scans available
• Many Application Areas
• Home Entertainment
• Construction Management
• Disaster Management
3. 3Challenge the future
General Introduction
• Problem 4D data: technical limitations of rendering system
• user‘s interests hard to find by traversing time-series data sets
• Goal: Visualisation algorithms for supportive user navigation
• Approaches:
• real-time rendered data traversal
• user-centred browsing
• navigation via visual summaries
Rendering & Navigation
Delfland dataset
1 time step
12.5 km * 10 km
takes up to 3 hours
to inspect in detail
User Interest:
(top) Landslide
probability;
(right) Door
Surveillance
4. 4Challenge the future
Research Statement
“I will search for algorithms and scalable representations that allow for
interaction, queries, and exploration of time-dependent point data .”
5. 5Challenge the future
Subtopics
1. Scalable Rendering and Visualisation of time-dependent
Point Sets
2. Efficiently Browsing through time-dependent Datasets
3. Navigation by Visual Summaries
6. 6Challenge the future
Scalable Rendering and Visualisation of
time-dependent Point Sets
• Interactive Rendering of large point sets already demanding
• additional, time-related challenges:
• just developing branch, virtually no data sets openly available
=> no available rendering approaches
• Rendering of multiple time steps faces technical challenges
3Di project:
currently
more than
12TB and
growing
7. 7Challenge the future
Scalable Rendering and Visualisation of
time-dependent Point Sets
• Goal: efficient rendering system
• displays massive point sets
• multiple time steps at the same time
• exploits visual and technical limitations
• Contribution:
• time-dependent LoD technique by continuous refinement
• temporal caching of point sets
8. 8Challenge the future
Subtopics
1. Scalable Rendering and Visualisation of time-dependent
Point Sets
2. Efficiently Browsing through time-dependent Datasets
3. Navigation by Visual Summaries
9. 9Challenge the future
Efficiently browsing through time-
dependent datasets
• Use Case: Surveillance of restricted areas
-> user-defined monitoring of areas
• Use Case: Natural Disaster Monitoring
-> changing demands and interests, depending on zoom level
10. 10Challenge the future
Efficiently browsing through
time-dependent datasets
Problems:
• browse while only showing user-specific interest
• remove/hide redundant data
• selection and interactive exploration of temporal data
• handling varying user interests according to user viewpoint
12. 12Challenge the future
Subtopics
1. Scalable Rendering and Visualisation of time-dependent
Point Sets
2. Efficiently Browsing through time-dependent Datasets
3. Navigation by Visual Summaries
13. 13Challenge the future
Navigation by Visual Summaries
• “Visual Summary”: visual and effective way to summarize
complex datasets
• various applications in entertainment, construction etc.
Creation
14. 14Challenge the future
Navigation by Visual Summaries
• party in Kinect-supervised house
• lots of objects (people) -> lots of events and interests
• goal: summarize the party to re-live it another day
Creation - Entertainment
15. 15Challenge the future
Navigation by Visual Summaries
• construction of building demands experts of different fields
• time constraints prevent meetings at construction site
• goal: summarize recent construction events for remote
planning
Creation – Construction
16. 16Challenge the future
Navigation by Visual Summaries
Problems:
• suitable representations
• spatio-temporal incoherence
• events wide-spread in space and time across the dataset
• suitable guidance to important events in the dataset
17. 18Challenge the future
Navigation by Visual Summaries
• user-driven interconnection to group objects of different steps
• intuitive user interface to regulate amount of spatial- and
temporal coherence
• test visual representations to determine the most suitable one
• compound visual summary as an album of impressions
Approach - Entertainment example
18. 19Challenge the future
Navigation by Visual Summaries
• focus on Guidance along events
• visual 4D tour
• 4D scene capture as interactive representation of a summary
Approach – Construction Example
19. 20Challenge the future
Conclusion
• focus: Visually navigating efficiently in time-dependent, multi-
dimensional, scanned data
• applications:
• Efficient Visual Surveillance
• Disaster Assessment with dynamically changing user interests
• Home Entertainment: Re-live a 3D-recorded party another day
• Visually guide construction processes
20. 21Challenge the future
Conclusion
• Necessary techniques:
• Real-time Rendering of Datasets
• Efficient Browsing through Datasets according to user interests
• Visual Summaries and their use as Guidance Method in Datasets
21. 22Challenge the future
The following 3 months
• finish paper “Interactive Rendering of Large-Scale, Geospatial
Data“ (PROM-4 Scientific Writing)
• generate artificial, temporal test dataset
Time-dependent Level-of-Detail technique by continuous refinement
• Construct tree structure via spatial subdivision
• Tree node refers to a list of time steps
• Each time-step stores hierarchical LoD tree
• low-resolution height maps or volumes per cell
• On traversal – continuous refinement:
• morph previously loaded points by rough estimate
• replace them gradually by newly loaded points
Height Difference
f(t-1, t)