effective optical flow estimation jan kamenický21.10.2011

14
Effective Optical Flow Estimation Jan Kamenický 21.10.2011

Upload: imogene-tucker

Post on 05-Jan-2016

219 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Effective Optical Flow Estimation

Jan Kamenický 21.10.2011

Page 2: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Motivation

Page 3: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Usage

• Motion detection• Object segmentation• Video encoding (compression)• Stereo disparity measurement

Page 4: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Optical flow equation

• Color constancy

• Taylor series

• Optical flow equation

Page 5: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Estimating optical flow

• Basic equation

• Local methods– Lucas & Kanade• flow field is locally constant (or affine)• least squares minimization• cannot handle interior parts of objects

Page 6: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Estimating optical flow

• Basic equation

• Global methods– Horn & Schunk

• more sensitive to noisedata term smoothing term

Page 7: Effective Optical Flow Estimation Jan Kamenický21.10.2011

• Data term– non-homogenous surface (shading, reflections)– non-flat scene / non-uniform lighting– spatial discontinuities

• Smoothness term– discontinuities (moving objects boundaries)

Problems

Page 8: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Estimating partial derivatives

• Discrete approximation by differences– forward, backward – not exact– use 2x2x2 cube in (x,y,t) space• compute the difference as an average of 4 adjacent first

order differences

– use larger support• e.g. [1, -8, 0, 8, -1]/12

i i+1

j

j+1

kk+1

Page 9: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Data term

• L2 norm

• L1 norm

• Many modifications– generalized Charbonnier

Page 10: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Regularization term

• Enforces smooth flow field• Similar norms can be used– L2, L1 (total variation), …

• Other possibilities– Laplacian instead of gradient

Page 11: Effective Optical Flow Estimation Jan Kamenický21.10.2011

Dealing with larger displacements

• Smoothing (blurring)– usually Gaussian kernel– decreases flow field accuracy

• Pyramidal approach– compute flow on down-sampled images– up-sample the flow to next level– compute the warping (using the optical flow)– repeat

Page 12: Effective Optical Flow Estimation Jan Kamenický21.10.2011

More optimizations

• Graduated non-convexity– iteratively move from convex energy function to

the more robust non-convex form

• Median filtering (5x5)– weighted modification

• More warping steps on one pyramid level

Page 13: Effective Optical Flow Estimation Jan Kamenický21.10.2011

OF methods comparison

• Optical flow estimation benchmark– http://vision.middlebury.edu/flow/

• Average end-point error

Page 14: Effective Optical Flow Estimation Jan Kamenický21.10.2011

References

• Main described method– D. Sun, S. Roth, M. J. Black: Secrets of Optical Flow

Estimation and Their Principles, CVPR 2010– http://www.cs.brown.edu/~black/code.html