66
Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

  • View
    300

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Segmentation and Perceptual Grouping

Kaniza

(Introduction to Computer Vision, 11.1.04)

Page 2: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 3: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 4: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 5: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 6: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

The image of this cube contradicts the optical image

Page 7: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 8: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 9: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 10: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 11: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Perceptual Organization

• Atomism, reductionism:– Perception is a process of decomposing an

image into its parts.– The whole is equal to the sum of its parts.

• Gestalt (Wertheimer, Köhler, Koffka 1912)– The whole is larger than the sum of its parts.

Page 12: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt: apparent motion

Page 13: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt: apparent motion

Page 14: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt Principles

• Proximity

Page 15: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt Principles

• Proximity• Proximity• Similarity

Page 16: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt Principles

• Proximity• Similarity

• Proximity• Similarity• Continuity

Page 17: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt Principles

• Closure• Proximity• Similarity• Continuity

Page 18: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt Principles

• Proximity• Similarity• Continuity

• Closure• Closure• Common Fate

Page 19: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Gestalt Principles

• Proximity• Similarity• Continuity

• Closure• Common Fate• Simplicity

• Closure• Common Fate

Page 20: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Mona Lisa

Page 21: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Mona Lisa

Page 22: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 23: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Smooth Completion

• Isotropic

• Smoothness

• Minimal curvature

• Extensibility

Elastica:

2min ( )k s ds

Page 24: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Elastica

• Scale invariant (Weiss, Bruckstein & Netravali)

• Approximation (Sharon, Brandt & Basri)

2 21 2 1 24( )

2min ( )l k s ds

Page 25: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

(Sharon, Brandt & Basri)

Page 26: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Hough Transform

Page 27: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Hough Transform

Page 28: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Curve Salience

Page 29: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Saliency Network

Encourage

• Length

• Low curvature

• Closure

(Shashua & Ullman)

Page 30: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Saliency Network(Shashua & Ullman)

Page 31: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Tensor Voting

• Every edge element votes to all its circular edge completions

• Vote attenuates with distance: e-αd

• Vote attenuates with curvature: e-βk

• Determine salience at every point using principal moments

(Guy & Medioni)

Page 32: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Tensor Voting(Guy & Medioni)

Page 33: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Stochastic Completion Field

• Random walk:

• In addition, a particle may die with probability:

2

cos

sin

(0, )

x

y

N

1/ re

(Mumford; Williams & Jacobs)

Page 34: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Stochastic Completion Fields

• Most probable path:

with

2

2

( )

1

21

log( 2 )

k s ds ds

r

(Mumford; Williams & Jacobs)

Page 35: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Stochastic Completion Fields(Mumford; Williams & Jacobs)

Page 36: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Stochastic Completion Fields(Mumford; Williams & Jacobs)

Page 37: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Stochastic Completion Fields(Mumford; Williams & Jacobs)

Page 38: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Shortest Path(Hu, Sakoda & Pavlidis)

Page 39: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Snakes

• Given a curve Г(s)=(x(s),y(s)), define:1

0

int

22 2

int 2

( ( ))

( ( ))

( , )

( ) ( )

image ext

image

E s ds

E s E E E

E I x y

E s ss s

(Kass, Witkin & Terzopolous)

Page 40: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Snakes: Curve Evolution

Page 41: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Snakes: Curve Evolution

Page 42: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Thresholding

Page 43: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Histogram

0 50 100 150 200 250

0

200

400

600

800

1000

1200

Page 44: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Thresholding

Page 45: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Thresholding

125

15699

Page 46: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Image Segmentation

Page 47: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Camouflage

Page 48: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 49: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 50: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 51: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 52: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 53: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Minimum Cut(Wu & Leahy)

Page 54: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Texture Examples

Page 55: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Filter Bank(Malik & Perona)

Page 56: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Normalized Cuts(Malik et al.)

Page 57: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Segmentation by Weighted Aggregation

A multiscale algorithm:• Optimizes a global measure• Returns a full hierarchy of segments• Linear complexity• Combines multiscale measurements:

– Texture– Boundary integrity

(Galun, Sharon, Brandt & Basri)

Page 58: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Segmentation by Weighted Aggregation(Galun, Sharon, Brandt & Basri)

Page 59: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Leopards

Page 60: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

And More…

Page 61: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)

Malik’s Ncuts

Page 62: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 63: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 64: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 65: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)
Page 66: Segmentation and Perceptual Grouping Kaniza (Introduction to Computer Vision, 11.1.04)