optical flow
DESCRIPTION
Optical Flow. Optical Flow. Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow. Optical Flow: Where do pixels move to?. Motion is a basic cue. Motion can be the only cue for segmentation. - PowerPoint PPT PresentationTRANSCRIPT
Optical Flow
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Optical Flow:Where do pixels move to?
Motion is a basic cueMotion can be the only cue for segmentation
Motion is a basic cue
Even impoverished motion data can elicit a strong percept
Applications tracking structure from motion motion segmentation stabilization compression mosaicing …
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Definition of optical flow
OPTICAL FLOW = apparent motion of brightness patterns
Ideally, the optical flow is the projection of the three-dimensional velocity vectors on the image
Caution required !
Two examples :
1. Uniform, rotating sphere
O.F. = 0
2. No motion, but changing lighting
O.F. 0
Mathematical formulation
I (x,y,t) = brightness at (x,y) at time t
Optical flow constraint equation :
0
tI
dtdy
yI
dtdx
xI
dtdI
),,(),,( tyxItttdtdyyt
dtdxxI
Brightness constancy assumption:
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
The aperture problem
0 tyx IvIuI
1 equation in 2 unknowns
dtdxu
dtdyv
yII x
yII y t
II t
Aperture Problem: Animation
The Aperture Problem
0
What is Optical Flow, anyway? Estimate of observed projected motion field Not always well defined! Compare:
Motion Field (or Scene Flow)projection of 3-D motion field
Normal Flowobserved tangent motion
Optical Flowapparent motion of the brightness pattern(hopefully equal to motion field)
Consider Barber pole illusion
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Horn & Schunck algorithm
Additional smoothness constraint :
,))()(( 2222 dxdyvvuue yxyxs besides OF constraint equation term
,)( 2 dxdyIvIuIe tyxc
minimize es+ec
Horn & Schunck
The Euler-Lagrange equations :
0
0
yx
yx
vvv
uuu
Fy
Fx
F
Fy
Fx
F
In our case ,
,)()()( 22222tyxyxyx IvIuIvvuuF
so the Euler-Lagrange equations are
,)(
,)(
ytyx
xtyx
IIvIuIv
IIvIuIu
2
2
2
2
yx
is the Laplacian operator
Horn & Schunck
Remarks :
1. Coupled PDEs solved using iterative methods and finite differences
2. More than two frames allow a better estimation of It
3. Information spreads from corner-type patterns
,)(
,)(
ytyx
xtyx
IIvIuIvtv
IIvIuIutu
Horn & Schunck, remarks
1. Errors at boundaries
2. Example of regularization (selection principle for the solution of ill-posed problems)
Results of an enhanced system
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
02),(
02),(
tyxy
tyxx
IvIuIIdv
vudE
IvIuIIdu
vudE
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow
Textured Motion http://
www.cs.ucla.edu/~wangyz/Research/wyzphdresearch.htm
Natural scenes contain rich stochastic motion patterns which are characterized by the movement of a large amount of particle andwave elements, such as falling snow, water waves, dancing grass, etc. We call these motion patterns "textured motion".