Progress on: Variable Viewpoint Reality Image Database
Overview of Presentation
VVR: Motivating Scenario
For example …
VVR Spectator Environment
Window of Opportunity
Progress at MIT
Visual Hull in 2D
Visual Hull: Segment
Visual Hull: Segment
Visual Hull: Segment
Visual Hull: Intersection
Idea in 2D: Visual Hull
Real Data: Tweety
Intersection of Frusta
New Apparatus
Current System
Visual Hull is very coarse …
Tomographic Reconstruction
Back-projection of image
intensities
Volume Render...
Poxels: An improvement to
tomography
Occlusion causes
disagreement
Initial agreement is not
enough…
Second pass uses information
about occlusion
Poxels Algorithm:
Definitions
Poxels: Model of
Transparency
Poxels Algorithm: Agreement
(Step 1)
Results…
From ICCV paper...
… additional results
Image Databases: Motivating
Scenario
But, the problem is very
hard…
We have made good
progress...
Search for cars?
Complex Feature
Representation
Image Database
Retrieval
NTT: Visit
1/7/99
Overview of IDB Meeting
Motivating Scenario
There is a very wide variety
of images...
Search for images containing
waterfalls?
Search for
cars?
What makes IDB hard?
Overview of IDB Meeting
Complex Feature
Representation
Slide 46
Slide 47
Resolution is reduced at
each step…
Not every feature is useful
for a query
Normalization of Signature
Space
Distance/Similarity Measure
Image Database Progress at
MIT
Slide 53
Slide 54
Slide 55
Slide 56
Conclusions