Extracting Textured Vertical Facades from Controlled CloseRange Imagery

1
Extracting Textured Vertical Facades from
Controlled Close-Range Imagery

• Satyan Coorg, Seth Teller, MIT Computer Graphics
  Group
• graphics.lcs.mit.edu

2
Novel system aspects

• Challenging datasets
• 1000s of real-world images (occlusion, foliage,
  clutter, lighting variations, shadows)
• No explicit feature correspondences
• Use of feature ensembles (e.g., edges)
• Scaling explicitly addressed
• A priori image grouping, spatial
  indexinggeodetic coordinates directly from
  sensor
• End-to-end system architecture
• Coarse structure, fine structure,aggregation,
  texture extraction

3
Dataset, Façade extraction
4
Recent results

• Aerial imagery, USGS terrain, sun model

5
Limitations
Navigation subsystem under development In
progress integration of GPS, inertial,
etc. Limited spatial extent (200 meters squared)
In progress acquisition through, above
MIT Exterior registration is semi-automated In
progress feature, ensemble correspondence Vertica
l facades only rooftops procedural Under way
polyhedral, quadric shapes Diffuse lighting,
diffuse surfaces Planned directional lighting
(a la Façade) Foliage removed via median stats
not modeled In progress foliage
segmentation, modeling
6
Please see our poster this afternoon!
And, for further information

• graphics.lcs.mit.edu
• graphics.lcs.mit.edu/seth
• graphics.lcs.mit.edu/city/city.html
• graphics.lcs.mit.edu/publications.html