CS201 Computer Vision Lect 08: SIFT Keypoint Detection John Magee - - PowerPoint PPT Presentation

▶
cs201 computer vision
SMART_READER_LITE
LIVE PREVIEW

CS201 Computer Vision Lect 08: SIFT Keypoint Detection John Magee - - PowerPoint PPT Presentation

Oct 18, 2023 248 likes •480 views

CS201 Computer Vision Lect 08: SIFT Keypoint Detection John Magee 23 Septermber 2014 Slides courtesy of Diane H. Theriault Question of the Day: How do we find repeatable, stable, scale- invariant points in images? SIFT Find

slide-1
SLIDE 1

CS201 – Computer Vision Lect 08: SIFT Keypoint Detection

John Magee 23 Septermber 2014

Slides courtesy of Diane H. Theriault

slide-2
SLIDE 2

Question of the Day:

  • How do we find repeatable, stable, scale-

invariant points in images?

slide-3
SLIDE 3
slide-4
SLIDE 4

SIFT

  • Find repeatable, scale-invariant points in images

(today)

  • Compute something about them
  • Use the thing you computed to perform matching
  • “Distinctive Image Features from Scale-Invariant

Keypoints” by David Lowe

  • Patented!
slide-5
SLIDE 5

How to find the same cat?

  • Imagine that we had a

library of cats

  • How could we find another

picture of the same cat in the library?

  • Look for the markings?
slide-6
SLIDE 6

How to find the same cat?

  • Imagine that we had a

library of cats

  • How could we find another

picture of the same cat in the library?

  • Look for the markings?
  • Which markings?
slide-7
SLIDE 7

How to find the same cat?

  • Imagine that we had a

library of cats

  • How could we find another

picture of the same cat in the library?

  • Look for the markings?
  • Which markings?
slide-8
SLIDE 8

How to find the same cat?

  • Imagine that we had a

library of cats

  • How could we find another

picture of the same cat in the library?

  • Look for the markings?
  • Which markings?
  • Which cat would be easier to

uniquely identify?

slide-9
SLIDE 9

Keypoints

  • Idea: Want to find points that are easy to

localize uniquely

This one Not this one

slide-10
SLIDE 10

Keypoints at Different Scales

  • Idea: Want to find scale-invariant points that

are easy to localize uniquely

slide-11
SLIDE 11

Gaussian

slide-12
SLIDE 12

Scale Space

  • Image convolved with Gaussians of different widths
slide-13
SLIDE 13

Scale Space Octaves

  • Every time the width of your Gaussian doubles, downsample

the image

slide-14
SLIDE 14

Derivative of a Gaussian

slide-15
SLIDE 15

Second Derivative of a Gaussian

slide-16
SLIDE 16

Laplacian of a Gaussian

  • Sum of spatial second derivatives
slide-17
SLIDE 17

Keypoints with Image Filtering

  • Perform image filtering by

convolving an image with a “filter”/”mask” / “kernel” to

  • btain a “result” / “response”
  • The value of the result will be

positive in regions of the image that “look like” the filter

  • What would a “dot” filter

look like? Image Filter

slide-18
SLIDE 18

Scale-space extrema

  • “Extrema” = local minimum
  • r maximum
  • Check 8 neighbors at a

particular scale

  • Check neighbors at scales

above and below

slide-19
SLIDE 19

Scale-space Extrema

slide-20
SLIDE 20

Difference of Gaussians

  • Approximation of the

Laplacian of a Gaussian

slide-21
SLIDE 21

Discussion Questions:

  • What does it mean for an image point to be

repeatable?

  • What are the properties of repeatable points?
slide-22
SLIDE 22

Keypoints with Image Filtering

  • What tool have we learned about for finding

places in the image that “look like” corners or dots?