EECS 274 Computer Vision Motion Estimation. Motion estimation Aligning images Estimate motion parameters Optical flow –Lucas-Kanade algorithm –Horn-Schunck.

EECS 274 Computer Vision

Motion Estimation

Motion estimation

• Aligning images• Estimate motion parameters• Optical flow

– Lucas-Kanade algorithm– Horn-Schunck algorithm

• Slides based on Szeliski’s lecture notes• Reading: FP Chapter 15, S Chapter 8

Why estimate visual motion?• Visual Motion can be annoying

– Camera instabilities, jitter– Measure it; remove it (stabilize)

• Visual Motion indicates dynamics in the scene– Moving objects, behavior– Track objects and analyze trajectories

• Visual Motion reveals spatial layout – Motion parallax

Classes of techniques

• Feature-based methods– Extract visual features (corners, textured areas) and track

them– Sparse motion fields, but possibly robust tracking– Suitable especially when image motion is large (10s of pixels)

• Direct methods– Directly recover image motion from spatio-temporal image

brightness variations– Global motion parameters directly recovered without an

intermediate feature motion calculation– Dense motion fields, but more sensitive to appearance

variations– Suitable for video and when image motion is small (< 10

pixels)

Optical flow vs. motion field

• Optical flow does not always correspond to motion field

• Optical flow is an approximation of the motion field• The error is small at points with high spatial gradient

under some simplifying assumptions

Patch matching

• How do we determine correspondences?

– block matching or SSD (sum squared differences)

Correlation and SSD

• For larger displacements, do template matching– Define a small area around a pixel as the

template– Match the template against each pixel within a

search area in next image– Use a match measure such as correlation,

normalized correlation, or sum-of-squares difference

– Choose the maximum (or minimum) as the match– Sub-pixel estimate (Lucas-Kanade)

Discrete search vs. gradient based• Consider image I translated by

• The discrete search method simply searches for the best estimate

• The gradient method linearizes the intensity function and solves for the estimate

21

,00

2

,1

1001

0

)),(),(),((

)),(),((),(

),(),(),(

),(),(

yxvvyuuxIyxI

vyuxIyxIvuE

yxyxIvyuxI

yxIyxI

yx

yx

00 ,vu

Brightness constancy

• Brightness Constancy Equation:

• Minimize :

• Linearize (assuming small (u,v))using Taylor series expansion:

image template theis

)),(),,((),(

0

10

I

yxvyyxuxIyxI

201 )),(),((),( yxIvyuxIvuE

),(),(),(),(),(),( 1111 yxvyxIyxuyxIyxIyxI yx

Estimating optical flow

• Assume the image intensity I is constant

Time = t Time = t+dt

dyydxx ,

yx,

),,(),,( 10 ttyyxxItyxI

Optical flow constraint

tt

Iy

y

Ix

x

ItyxItyxI

ttyyxxItyxI

),,(),,(

),,(),,(

v

2)(),(

;,

,

0

0let and ,0

tyx

Ttyx

tyx

tyx

IvIuIvuE

AIIv

uII

IvIuI

IvIuI

tt

I

t

y

y

I

t

x

x

I

bubu

u

),( yx II

Lucas-Kanade algorithm

Minimizing

Assume a single velocity for all pixels within an image patch

LHS: sum of the 2x2 outer product of the gradient vector

yx

tyx IvyxIuyxIvuE,

2)),(),((),(

tT

ty

tx

yyx

yxx

TT

IIuII

II

II

v

u

III

III

AAA

A

)(

)(

2

2

1

2

bu

bu

Hessian

Matrix form

)()(

)(,)()(

)(

)()()(

)()()(

1

1111

21

211

21

ioii

iii

iii

iioii

iioi

IIe

y

I

x

II

e

II

IIE

xux

uxuxuxJ

uuxJ

xuuxJux

xuuxuu

Matrix form

)()(

)()(

)(,)()(

,

)(

)()(

01

1

1111

2

2

1

11

ii

ioii

iii

ty

tx

yyx

yxx

iii

iii

T

IIe

y

I

x

II

II

II

III

IIIA

uxe

A

A

xJuxJ

xux

uxuxuxJ

b

Jb

uxJuxJ

bu

for computational efficiency

Computing gradients in X-Y-T

k

k+1

i i+1

j

j+1

time

x

y

)](

)[(4

1

1,1,,1,1,,,,

1,1,1,1,11,,1,,1

kjikjikjikji

kjikjikjikjix

IIII

IIIIx

I

likewise for Iy and It

Local patch analysis

• How certain are the motion estimates?

The aperture problem

• Algorithm: At each pixel compute by solving

• A is singular if all gradient vectors point in the same direction• e.g., along an edge• of course, trivially singular if the summation is over a single pixel or there is no texture• i.e., only normal flow is available (aperture problem)

• Corners and textured areas are OK

ty

txT

II

IIIIA band,Let

u bu A

SSD surface – textured area

SSD surface – edge

SSD – homogeneous area

Iterative refinement

• Estimate velocity at each pixel using one iteration of Lucas and Kanade estimation

• Warp one image toward the other using the estimated flow field(easier said than done)

• Refine estimate by repeating the process

Optical flow: iterative estimation

xx0

Initial guess:

Estimate:

estimate update

(using d for displacement here instead of u)

Optical flow: iterative estimation

xx0

estimate update

Initial guess:

Estimate:

Optical flow: iterative estimation

xx0

Initial guess:

Estimate:

Initial guess:

Estimate:

estimate update

Optical flow: iterative estimation

xx0

Optical flow: iterative estimation• Some implementation issues:

– Warping is not easy (ensure that errors in warping are smaller than the estimate refinement)

– Warp one image, take derivatives of the other so you don’t need to re-compute the gradient after each iteration

– Often useful to low-pass filter the images before motion estimation (for better derivative estimation, and linear approximations to image intensity)

Error functions

• Robust error function

• Spatially varying weights

• Bias and gain: images taken with different exposure

• Correlation (and normalized cross correlation)

22

2

/1)()),()(()(

ax

xxIIE

ii

xuxu

i

iiii IIwwE 210 )()()()()( xuxuxxu

2)()1()()(

bias theis andgain theis ,)()1()(

iii IIE

II

xuxu

xux

i

ii IIE )()()( uxxu

Horn-Schunck algorithm

• Global method with smoothness constraint to solve aperture problem

• Minimize a global energy functional with calculus of variations

dxdyvuIvIuIE tyx 2222

functionenergy theof integrand theis where

0

0

L

v

L

yv

L

xv

L

u

L

yu

L

xu

L

yx

yx

Horn-Schunck algorithm

operator Laplace theis where

0)(

0)(

2

2

2

2

2

2

yx

vIvIuII

uIvIuII

tyxy

tyxx

tyyyx

txyxx

IIvvIuII

IIuvIIuI

yxuyxuyxu

222

222

)(

)(

),(),(),(

See Robot Vision by Horn for details

Horn-Schunck algorithm

• Iterative scheme

• Yields high density flow• Fill in missing information in the

homogenous regions

• More sensitive to noise than local methods

2221

2221

)(

)(

yx

tk

yk

xykk

yx

tk

yk

xxkk

II

IvIuIIvv

II

IvIuIIuu

u

v

),( yx II

),( vu

),( vu

Optical flow: aliasing

Temporal aliasing causes ambiguities in optical flow because images can have many pixels with the same intensity.

I.e., how do we know which ‘correspondence’ is correct?

nearest match is correct (no aliasing)

nearest match is incorrect (aliasing)

To overcome aliasing: coarse-to-fine estimationcoarse-to-fine estimation.

actual shift

estimated shift

image Iimage J

aJwwarp refine

a

aΔ+

Pyramid of image J Pyramid of image I

image Iimage J

Coarse-to-fine estimation

u=10 pixels

u=5 pixels

u=2.5 pixels

u=1.25 pixels

J Jw Iwarp refine

ina

a

+

J Jw Iwarp refine

a

a+

J

pyramid construction

J Jw Iwarp refine

a+

I

pyramid construction

outa

Coarse-to-fine estimation

Global (parametric) motion models• 2D Models:• Affine• Quadratic• Planar projective transform (Homography)

• 3D Models:• Instantaneous camera motion models • Homography+epipole• Plane+Parallax

Motion models

Translation

2 unknowns

Affine

6 unknowns

Perspective

8 unknowns

3D rotation

3 unknowns

Example: affine motion

• Substituting into the B.C. equation

Each pixel provides 1 linear constraint in 6 global unknowns

Least square minimization (over all pixels):

2 tyx IyaxaaIyaxaaIaErr )()()( 654321

0)()( 654321 tyx IyaxaaIyaxaaI

0 tyx IvIuI

yaxaayxv

yaxaayxu

654

321

),(

),(

Parametric motion

)'('

)'()'(

)'(

)()'()'(

)());('(

map encecorrespondor fieldmotion :);('

11

1

21

2011

201

ii

ii

iii

iiii

iii

II

e

II

IIE

xp

xx

pxJ

pxJ

xpxJx

xppxx

pxx

i.e., the product of image gradient with Jacobian of the correspondence field

Parametric motion

)'()(

)()'()'()(

1

11

'

''

iT

ii

iT

iT

iiT

ii

T

Ieb

IIA

i

ii

xxJ

xJxxxJ

x

xx

the expressions inside the brackets are the same as the cases for simpler translation motion case

Other 2D Motion Models

Quadratic – instantaneous approximation to planar motion 2

87654

82

7321

yqxyqyqxqqv

xyqxqyqxqqu

yyvxxu

yhxhh

yhxhhy

yhxhh

yhxhhx

','

and

'

'

987

654

987

321

Projective – exact planar motion

3D Motion Models

ZxTTxxyyv

ZxTTyxxyu

ZYZYX

ZXZYX

)()1(

)()1(2

2

yyvxxu

thyhxh

thyhxhy

thyhxh

thyhxhx

',' :and

'

'

3987

1654

3987

1321

)(1

)(1

233

133

tytt

xyv

txtt

xxu

w

w

Local Parameter:

ZYXZYX TTT ,,,,,

),( yxZ

Instantaneous camera motion:

Global parameters:

Global parameters: 32191 ,,,,, ttthh

),( yx

Homography+Epipole

Local Parameter:

Residual Planar Parallax Motion

Global parameters: 321 ,, ttt

),( yxLocal Parameter:

Shi-Tomasi feature tracker

1. Find good features (min eigenvalue of 22 Hessian)

2. Use Lucas-Kanade to track with pure translation

3. Use affine registration with first feature patch

4. Terminate tracks whose dissimilarity gets too large

5. Start new tracks when needed

Tracking results

Tracking - dissimilarity

Tracking results

Correlation window size

• Small windows lead to more false matches• Large windows are better this way, but…

– Neighboring flow vectors will be more correlated (since the template windows have more in common)

– Flow resolution also lower (same reason)– More expensive to compute

• Small windows are good for local search:more detailed and less smooth (noisy?)

• Large windows good for global search:less detailed and smoother

Robust estimation

Noise distributions are often non-Gaussian, having much heavier tails. Noise samples from the tails are called outliers.

• Sources of outliers (multiple motions):– specularities / highlights– jpeg artifacts / interlacing / motion blur– multiple motions (occlusion boundaries, transparency)velocity spacevelocity space

u1

u2

++

How to handle background and foreground motion

Robust estimation

Standard Least Squares Estimation allows too much influence for outlying points

)()

)()(

)()(

2

mxx

x

mxx

xmE

i

ii

ii

( Influence

Robust estimation

Robust gradient constraint

Robust SSD

),(),(),(

),(

ssssd

tsysxssd

vyuxIyxIvuE

IvIuIvuE

Robust estimation

Problem: Least-squares estimators penalize deviations between data & model with quadratic error fn (extremely sensitive to outliers)

error penalty function influence function

Redescending error functions (e.g., Geman-McClure) help to reduce the influence of outlying measurements.

error penalty function influence function

2)( 2

)()(

2

2

);(

s

s 22 )(

2);(

s

ss

Performance evaluation

• See Baker et al. “A Database and Evaluation Methodology for Optical Flow”, ICCV 2007

• Algorithms:• Pyramid LK: OpenCV-based implementation of

Lucas-Kanade on a Gaussian pyramid• Black and Anandan: Author’s implementation

• Bruhn et al.: Our implementation• MediaPlayerTM: Code used for video frame-rate

upsampling in Microsoft MediaPlayer• Zitnick et al.: Author’s implementation

Performance evaluation

• Difficulty: Data substantially more challenging than Yosemite

• Diversity: Substantial variation in difficulty across the various datasets

• Motion GT vs Interpolation: Best algorithms for one are not the best for the other

• Comparison with Stereo: Performance of existing flow algorithms appears weak

Motion representations

• How can we describe this scene?

Block-based motion prediction• Break image up into square blocks• Estimate translation for each block• Use this to predict next frame, code

difference (MPEG-2)

Layered motion

• Break image sequence up into “layers”:

=

• Describe each layer’s motion

Layered motion

• Advantages:• can represent occlusions / disocclusions• each layer’s motion can be smooth• video segmentation for semantic

processing• Difficulties:• how do we determine the correct number?• how do we assign pixels?• how do we model the motion?

Layers for video summarization

Background modeling (MPEG-4)• Convert masked images into a

background sprite for layered video coding

+ + + + + +

=

What are layers?

• Intensities• Velocities• Layers• Alpha matting• Sprites• Wang and Adelson,

“Representing moving images with layers”, IEEE Transactions on Image Processing, 1994

How do we form them?

How do we estimate the layers?1. compute coarse-to-fine flow2. estimate affine motion in blocks

(regression)3. cluster with k-means4. assign pixels to best fitting affine region5. re-estimate affine motions in each

region…

Layer synthesis

• For each layer:• stabilize the sequence with the affine

motion• compute median value at each pixel

• Determine occlusion relationships

Results

Applications

• Motion analysis• Video coding• Morphing• Video denoising• Video stabilization • Medical image registration• Frame interpolation