project 3 questions?

61

Upload: nellie

Post on 15-Jan-2016

40 views

Category:

Documents


0 download

DESCRIPTION

Project 3 questions?. 10/21/11. Interest Points and Instance Recognition. Computer Vision CS 143, Brown James Hays. Many slides from Kristen Grauman and Derek Hoiem. Last time. Detecting corner-like points in an image. Today. Local invariant features Detection of interest points - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Project 3 questions?
Page 2: Project 3 questions?

Project 3 questions?

Page 3: Project 3 questions?

Interest Points and Instance Recognition

Computer VisionCS 143, Brown

James Hays

10/21/11

Many slides from Kristen Grauman and Derek Hoiem

Page 4: Project 3 questions?

Last time

• Detecting corner-like points in an image

Page 5: Project 3 questions?

Today

• Local invariant features– Detection of interest points

• (Harris corner detection)• Scale invariant blob detection: LoG

– Description of local patches• SIFT pipeline for invariant local features

Page 6: Project 3 questions?

Local invariant features: outline1) Detection: Identify the interest

points

2) Description: Extract vector feature descriptor surrounding each interest point.

3) Matching: Determine correspondence between descriptors in two views

],,[ )1()1(11 dxx x

],,[ )2()2(12 dxx x

Kristen Grauman

Page 7: Project 3 questions?

Goal: interest operator repeatability

• We want to detect (at least some of) the same points in both images.

• Yet we have to be able to run the detection procedure independently per image.

No chance to find true matches!

Page 8: Project 3 questions?

Goal: descriptor distinctiveness

• We want to be able to reliably determine which point goes with which.

• Must provide some invariance to geometric and photometric differences between the two views.

?

Page 9: Project 3 questions?

Local features: main components1) Detection: Identify the interest

points

2) Description:Extract vector feature descriptor surrounding each interest point.

3) Matching: Determine correspondence between descriptors in two views

Page 10: Project 3 questions?

Since M is symmetric, we have TXXM

2

1

0

0

iii xMx

The eigenvalues of M reveal the amount of intensity change in the two principal orthogonal gradient directions in the window.

Recall: Corners as distinctive interest points

Page 11: Project 3 questions?

“flat” region1 and 2 are small;

“edge”:1 >> 2

2 >> 1

“corner”:1 and 2 are large, 1 ~ 2;

One way to score the cornerness:

Recall: Corners as distinctive interest points

Page 12: Project 3 questions?

Slide: Derek Hoiem

Harris Detector [Harris88]

• Second moment matrix(autocorrelation matrix)

)()(

)()()(),( 2

2

DyDyx

DyxDxIDI III

IIIg

12

1. Image derivatives

2. Square of derivatives

3. Gaussian filter g(I)

Ix Iy

Ix2 Iy

2 IxIy

g(Ix2) g(Iy

2) g(IxIy)

222222 )]()([)]([)()( yxyxyx IgIgIIgIgIg

])),([trace()],(det[ 2DIDIhar

4. Cornerness function – both eigenvalues are strong

har5. Non-maxima suppression

1 2

1 2

det

trace

M

M

Page 13: Project 3 questions?

Properties of the Harris corner detector

Rotation invariant?

Scale invariant?

TXXM

2

1

0

0

Yes

Page 14: Project 3 questions?

Properties of the Harris corner detector

Rotation invariant?

Scale invariant?

All points will be classified as edges

Corner !

Yes

No

Page 15: Project 3 questions?

Scale invariant interest points

How can we independently select interest points in each image, such that the detections are repeatable across different scales?

Page 16: Project 3 questions?

Automatic scale selection

Intuition: •Find scale that gives local maxima of some function f in both position and scale.

f

region size

Image 1f

region size

Image 2

s1 s2

Page 17: Project 3 questions?

What can be the “signature” function?

Page 18: Project 3 questions?

Blob detection in 2D

Laplacian of Gaussian: Circularly symmetric operator for blob detection in 2D

2

2

2

22

y

g

x

gg

Page 19: Project 3 questions?

Blob detection in 2D: scale selection

Laplacian-of-Gaussian = “blob” detector2

2

2

22

y

g

x

gg

fil

ter

scal

es

img1 img2 img3Bastian Leibe

Page 20: Project 3 questions?

Blob detection in 2D

We define the characteristic scale as the scale that produces peak of Laplacian response

characteristic scale

Slide credit: Lana Lazebnik

Page 21: Project 3 questions?

Example

Original image at ¾ the size

Kristen Grauman

Page 22: Project 3 questions?

Original image at ¾ the size

Kristen Grauman

Page 23: Project 3 questions?

Kristen Grauman

Page 24: Project 3 questions?

Kristen Grauman

Page 25: Project 3 questions?

Kristen Grauman

Page 26: Project 3 questions?

Kristen Grauman

Page 27: Project 3 questions?

)()( yyxx LL

List of (x, y, σ)

scale

Scale invariant interest points

Interest points are local maxima in both position and scale.

Squared filter response maps

Page 28: Project 3 questions?

Scale-space blob detector: Example

Image credit: Lana Lazebnik

Page 29: Project 3 questions?

We can approximate the Laplacian with a difference of Gaussians; more efficient to implement.

2 ( , , ) ( , , )xx yyL G x y G x y

( , , ) ( , , )DoG G x y k G x y

(Laplacian)

(Difference of Gaussians)

Technical detail

Page 30: Project 3 questions?

Maximally Stable Extremal Regions [Matas ‘02]

• Based on Watershed segmentation algorithm• Select regions that stay stable over a large

parameter range

K. Grauman, B. Leibe

Page 31: Project 3 questions?

Example Results: MSER

34 K. Grauman, B. Leibe

Page 32: Project 3 questions?

Comparison

LoG

MSER

Harris

Page 33: Project 3 questions?

Comparison of Keypoint Detectors

Tuytelaars Mikolajczyk 2008

Page 34: Project 3 questions?

Choosing a detector

• What do you want it for?– Precise localization in x-y: Harris– Good localization in scale: Difference of Gaussian– Flexible region shape: MSER

• Best choice often application dependent– Harris-/Hessian-Laplace/DoG work well for many natural categories– MSER works well for buildings and printed things

• Why choose?– Get more points with more detectors

• There have been extensive evaluations/comparisons– [Mikolajczyk et al., IJCV’05, PAMI’05]– All detectors/descriptors shown here work well

Page 35: Project 3 questions?

• For most local feature detectors, executables are available online:– http://robots.ox.ac.uk/~vgg/research/affine– http://www.cs.ubc.ca/~lowe/keypoints/– http://www.vision.ee.ethz.ch/~surf

K. Grauman, B. Leibe

Page 36: Project 3 questions?

Local features: main components1) Detection: Identify the interest

points

2) Description:Extract vector feature descriptor surrounding each interest point.

3) Matching: Determine correspondence between descriptors in two views

],,[ )1()1(11 dxx x

],,[ )2()2(12 dxx x

Page 37: Project 3 questions?

Geometric transformations

e.g. scale, translation, rotation

Page 38: Project 3 questions?

Photometric transformations

Figure from T. Tuytelaars ECCV 2006 tutorial

Page 39: Project 3 questions?

Raw patches as local descriptors

The simplest way to describe the neighborhood around an interest point is to write down the list of intensities to form a feature vector.

But this is very sensitive to even small shifts, rotations.

Page 40: Project 3 questions?

SIFT descriptor [Lowe 2004]

• Use histograms to bin pixels within sub-patches according to their orientation.

0 2

Why subpatches?

Why does SIFT have some illumination invariance?

Page 41: Project 3 questions?

CSE 576: Computer Vision

Making descriptor rotation invariant

Image from Matthew Brown

• Rotate patch according to its dominant gradient orientation

• This puts the patches into a canonical orientation.

Page 42: Project 3 questions?

• Extraordinarily robust matching technique• Can handle changes in viewpoint

• Up to about 60 degree out of plane rotation

• Can handle significant changes in illumination• Sometimes even day vs. night (below)

• Fast and efficient—can run in real time• Lots of code available

• http://people.csail.mit.edu/albert/ladypack/wiki/index.php/Known_implementations_of_SIFT

Steve Seitz

SIFT descriptor [Lowe 2004]

Page 43: Project 3 questions?

Example

NASA Mars Rover images

Page 44: Project 3 questions?

NASA Mars Rover imageswith SIFT feature matchesFigure by Noah Snavely

Example

Page 45: Project 3 questions?

SIFT properties

• Invariant to– Scale – Rotation

• Partially invariant to– Illumination changes– Camera viewpoint– Occlusion, clutter

Page 46: Project 3 questions?

Local features: main components1) Detection: Identify the interest

points

2) Description:Extract vector feature descriptor surrounding each interest point.

3) Matching: Determine correspondence between descriptors in two views

Page 47: Project 3 questions?

Matching local features

Kristen Grauman

Page 48: Project 3 questions?

Matching local features

?

To generate candidate matches, find patches that have the most similar appearance (e.g., lowest SSD)

Simplest approach: compare them all, take the closest (or closest k, or within a thresholded distance)

Image 1 Image 2

Kristen Grauman

Page 49: Project 3 questions?

Ambiguous matches

At what SSD value do we have a good match?

To add robustness to matching, can consider ratio : distance to best match / distance to second best match

If low, first match looks good.

If high, could be ambiguous match.

Image 1 Image 2

? ? ? ?

Kristen Grauman

Page 50: Project 3 questions?

Matching SIFT Descriptors• Nearest neighbor (Euclidean distance)• Threshold ratio of nearest to 2nd nearest descriptor

Lowe IJCV 2004

Page 51: Project 3 questions?

Recap: robust feature-based alignment

Source: L. Lazebnik

Page 52: Project 3 questions?

Recap: robust feature-based alignment

• Extract features

Source: L. Lazebnik

Page 53: Project 3 questions?

Recap: robust feature-based alignment

• Extract features• Compute putative matches

Source: L. Lazebnik

Page 54: Project 3 questions?

Recap: robust feature-based alignment

• Extract features• Compute putative matches• Loop:

• Hypothesize transformation T (small group of putative matches that are related by T)

Source: L. Lazebnik

Page 55: Project 3 questions?

Recap: robust feature-based alignment

• Extract features• Compute putative matches• Loop:

• Hypothesize transformation T (small group of putative matches that are related by T)

• Verify transformation (search for other matches consistent with T)

Source: L. Lazebnik

Page 56: Project 3 questions?

Recap: robust feature-based alignment

• Extract features• Compute putative matches• Loop:

• Hypothesize transformation T (small group of putative matches that are related by T)

• Verify transformation (search for other matches consistent with T)

Source: L. Lazebnik

Page 57: Project 3 questions?

Applications of local invariant features

• Wide baseline stereo• Motion tracking• Panoramas• Mobile robot navigation• 3D reconstruction• Recognition• …

Page 58: Project 3 questions?

Automatic mosaicing

http://www.cs.ubc.ca/~mbrown/autostitch/autostitch.html

Page 59: Project 3 questions?

Wide baseline stereo

[Image from T. Tuytelaars ECCV 2006 tutorial]

Page 60: Project 3 questions?

Recognition of specific objects, scenes

Rothganger et al. 2003 Lowe 2002

Schmid and Mohr 1997 Sivic and Zisserman, 2003

Kristen Grauman

Page 61: Project 3 questions?

Summary

• Interest point detection– Harris corner detector– Laplacian of Gaussian, automatic scale selection

• Invariant descriptors– Rotation according to dominant gradient direction– Histograms for robustness to small shifts and

translations (SIFT descriptor)