Presentation is loading. Please wait.

Presentation is loading. Please wait.

CS332 Visual Processing Department of Computer Science Wellesley College Analysis of Motion Recovering 3-D structure from motion.

Similar presentations


Presentation on theme: "CS332 Visual Processing Department of Computer Science Wellesley College Analysis of Motion Recovering 3-D structure from motion."— Presentation transcript:

1 CS332 Visual Processing Department of Computer Science Wellesley College Analysis of Motion Recovering 3-D structure from motion

2 1-2 Recovering 3D structure from motion

3 1-3 Ambiguity of 3D recovery birds’ eye views We need additional constraint to recover 3D structure uniquely “rigidity constraint”

4 1-4 Image projections Z X perspective projection image plane Z X orthographic projection image plane (X, Y, Z)  (X, Y) (X, Y, Z)  (X/Z, Y/Z)  only relative depth  requires object rotation  only scaled depth  requires translation of observer relative to scene

5 1-5 Using the rigidity constraint Ullman: Given 3 distinct orthographic views of 4 non-coplanar points in motion, if there exists a rigid 3-D structure consistent with these views, then this structure is unique. view 1view 2view 3 2D positions of points in 3 views set of equations to solve 3D structure of points caveat: depth reversals

6 1-6 What is needed to compute a unique rigid 3D structure? 4 points 3 views 7 points 2 views 5 points 1-1/2 views 5 points 1 view orthographic projection perspective projection correspondence velocity field Sample results: * * Ullman Rigidity constraint alone is sufficient to compute 3D structure from motion, BUT…

7 1-7 Human recovery of 3D structure from motion:  Needs extended time to obtain an accurate 3D structure, and derivation is not all-or-none  Can cope with significant deviations from rigidity  Integrates multiple sources of information These factors motivated the design of Ullman’s incremental rigidity scheme

8 1-8 Incremental Rigidity Scheme x y image (x 1 y 1 z 1 ) (x 2 y 2 z 2 ) (x 3 y 3 z 3 ) (x 1 ' y 1 ' ??) (x 2 ' y 2 ' ??) (x 3 ' y 3 ' ??) depth: Z initially, Z=0 at all points Find new 3D model that maximizes rigidity Compute new Z values that minimize change in 3D structure

9 1-9 Bird’s eye view: depth x image current model new image Find new Z i that minimize Σ (L ij – l ij ) 2 /L ij 3 L ij l ij Z

10 1-10 Incremental rigidity scheme: results bird’s eye views new 3D model after every 10° rotation true structure computed 3D model builds 3D model incrementally over extended time

11 1-11 Spontaneous depth reversal true structure computed 3D model

12 1-12 Human recovery of 3D structure from motion  2-3 points sufficient  requires extended time to build up accurate percept of 3-D structure Anderson & colleagues


Download ppt "CS332 Visual Processing Department of Computer Science Wellesley College Analysis of Motion Recovering 3-D structure from motion."

Similar presentations


Ads by Google