lecture 2: color, filtering & edgesfergus/teaching/vision_2012/2_color_filtering... · color...
TRANSCRIPT
![Page 1: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/1.jpg)
Lecture 2:
Color, Filtering & Edges
Slides: S. Lazebnik, S. Seitz, W. Freeman, F. Durand, D. Forsyth, D. Lowe, B. Wandell, S.Palmer, K. Grauman
![Page 2: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/2.jpg)
Color
![Page 3: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/3.jpg)
What is color?
•
•
![Page 4: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/4.jpg)
Color Camera Sensor
http://www.photoaxe.com/wp-content/uploads/2007/04/camera-sensor.jpg
![Page 5: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/5.jpg)
Overview of Color
•
•
•
•
![Page 6: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/6.jpg)
Electromagnetic Spectrum
http://www.yorku.ca/eye/photopik.htm
![Page 7: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/7.jpg)
Why do we see light of these wavelengths?
© Stephen E. Palmer, 2002
…because that’s where the
Sun radiates EM energy
Visible Light
![Page 8: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/8.jpg)
The Physics of Light
Any source of light can be completely described
physically by its spectrum: the amount of energy emitted
(per time unit) at each wavelength 400 - 700 nm.
© Stephen E. Palmer, 2002
400 500 600 700
Wavelength (nm.)
# Photons(per ms.)
Relative
spectral
power
![Page 9: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/9.jpg)
The Physics of Light
.
# P
ho
ton
s
D. Normal Daylight
Wavelength (nm.)
B. Gallium Phosphide Crystal
400 500 600 700
# P
ho
ton
s
Wavelength (nm.)
A. Ruby Laser
400 500 600 700
400 500 600 700
# P
ho
ton
s
C. Tungsten Lightbulb
400 500 600 700
# P
ho
ton
s
Some examples of the spectra of light sources
© Stephen E. Palmer, 2002
Rel. p
ow
er
Rel. p
ow
er
Rel. p
ow
er
Rel. p
ow
er
![Page 10: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/10.jpg)
The Physics of Light
Some examples of the reflectance spectra of surfaces
Wavelength (nm)
% L
ight
Reflecte
d
Red
400 700
Yellow
400 700
Blue
400 700
Purple
400 700
© Stephen E. Palmer, 2002
![Page 11: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/11.jpg)
Interaction of light and surfaces
•
•
–
–
From Foundation of Vision by Brian Wandell, Sinauer Associates, 1995
![Page 12: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/12.jpg)
Interaction of light and surfaces
•
Olafur Eliasson, Room for one colorSlide by S. Lazebnik
![Page 13: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/13.jpg)
Overview of Color
•
•
•
•
![Page 14: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/14.jpg)
The Eye
The human eye is a camera!• Iris - colored annulus with radial muscles
• Pupil - the hole (aperture) whose size is controlled by the iris
• What‟s the “film”?– photoreceptor cells (rods and cones) in the retina
Slide by Steve Seitz
![Page 15: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/15.jpg)
The Retina
Cross-section of eye
Ganglion cell layer
Bipolar cell layer
Receptor layer
Pigmentedepithelium
Ganglion axons
Cross section of retina
![Page 16: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/16.jpg)
Retina up-close
Light
![Page 17: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/17.jpg)
© Stephen E. Palmer, 2002
Cones
cone-shaped
less sensitive
operate in high light
color vision
Two types of light-sensitive receptors
Rods
rod-shaped
highly sensitive
operate at night
gray-scale vision
![Page 18: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/18.jpg)
Rod / Cone sensitivity
The famous sock-matching problem…
![Page 19: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/19.jpg)
© Stephen E. Palmer, 2002
.
400 450 500 550 600 650
RE
LA
TIV
E A
BS
OR
BA
NC
E (
%)
WAVELENGTH (nm.)
100
50
440
S
530 560 nm.
M L
Three kinds of cones:
Physiology of Color Vision
• Why are M and L cones so close?
• Are are there 3?
![Page 20: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/20.jpg)
Color perception
Rods and cones act as filters on the spectrum• To get the output of a filter, multiply its response curve by the
spectrum, integrate over all wavelengths
– Each cone yields one number
S
M L
Wavelength
Power
• Q: How can we represent an entire spectrum with 3 numbers?
• A: We can‟t! Most of the information is lost.
– As a result, two different spectra may appear indistinguishable
» such spectra are known as metamersSlide by Steve Seitz
![Page 21: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/21.jpg)
Spectra of some real-world surfaces
metamers
![Page 22: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/22.jpg)
Standardizing color experience
• We would like to understand which spectra
produce the same color sensation in people
under similar viewing conditions
• Color matching experiments
Foundations of Vision, by Brian Wandell, Sinauer Assoc., 1995
![Page 23: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/23.jpg)
Color matching experiment 1
Source: W. Freeman
![Page 24: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/24.jpg)
Color matching experiment 1
p1 p2 p3
Source: W. Freeman
![Page 25: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/25.jpg)
Color matching experiment 1
p1 p2 p3
Source: W. Freeman
![Page 26: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/26.jpg)
Color matching experiment 1
p1 p2 p3
The primary color
amounts needed for a
match
Source: W. Freeman
![Page 27: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/27.jpg)
Color matching experiment 2
Source: W. Freeman
![Page 28: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/28.jpg)
Color matching experiment 2
p1 p2 p3
Source: W. Freeman
![Page 29: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/29.jpg)
Color matching experiment 2
p1 p2 p3
Source: W. Freeman
![Page 30: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/30.jpg)
Color matching experiment 2
p1 p2 p3p1 p2 p3
We say a “negative”
amount of p2 was
needed to make the
match, because we
added it to the test
color’s side.
The primary color
amounts needed for a
match:
p1 p2 p3
Source: W. Freeman
![Page 31: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/31.jpg)
Trichromacy
•
–
•
–
•
–
–
![Page 32: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/32.jpg)
Grassman’s Laws
•
•
–
•
–
•
–
![Page 33: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/33.jpg)
Overview of Color
•
•
•
•
![Page 34: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/34.jpg)
Linear color spaces
•
•
mixing two lights producescolors that lie along a straight
line in color space
mixing three lights produces colors that lie within the triangle
they define in color space
![Page 35: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/35.jpg)
How to compute the weights of the primaries
to match any spectral signal
• Matching functions: the amount of each primary
needed to match a monochromatic light source at
each wavelength
p1 p2 p3
Source: W. Freeman
![Page 36: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/36.jpg)
RGB space
• Primaries are monochromatic lights (for monitors, they correspond to the three types of phosphors)
• Subtractive matching required for some wavelengths
RGB matching functionsRGB primaries
![Page 37: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/37.jpg)
How to compute the weights of the primaries
to match any spectral signal
• Let c(λ) be one of the matching functions, and let t(λ) be
the spectrum of the signal. Then the weight of the
corresponding primary needed to match t is
dtcw )()(
λ
Matching functions, c(λ)
Signal to be matched, t(λ)
![Page 38: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/38.jpg)
Linear color spaces: CIE XYZ
• Primaries are imaginary, but matching functions are everywhere positive
• The Y parameter corresponds to brightness or luminance of a color
• 2D visualization: draw (x,y), where x = X/(X+Y+Z), y = Y/(X+Y+Z)
Matching functions
http://en.wikipedia.org/wiki/CIE_1931_color_space
![Page 39: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/39.jpg)
Nonlinear color spaces: HSV
• Perceptually meaningful dimensions:
Hue, Saturation, Value (Intensity)
• RGB cube on its vertex
![Page 40: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/40.jpg)
Useful reference
Stephen E. Palmer, Vision Science: Photons
to Phenomenology, MIT Press, 1999
![Page 41: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/41.jpg)
Overview of Color
•
•
•
•
![Page 42: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/42.jpg)
White balance
• When looking at a picture on screen or print, we adapt to
the illuminant of the room, not to that of the scene in the
picture
• When the white balance is not correct, the picture will
have an unnatural color “cast”
http://www.cambridgeincolour.com/tutorials/white-balance.htm
incorrect white balance correct white balance
![Page 43: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/43.jpg)
White balance
• Film cameras: • Different types of film or different filters for different
illumination conditions
• Digital cameras: • Automatic white balance
• White balance settings corresponding to
several common illuminants
• Custom white balance using a reference
object
http://www.cambridgeincolour.com/tutorials/white-balance.htm Slide: F. Durand
![Page 44: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/44.jpg)
White balance
• Von Kries adaptation• Multiply each channel by a gain factor
• A more general transformation would correspond to an arbitrary 3x3
matrix
Slide: F. Durand
![Page 45: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/45.jpg)
White balance
• Von Kries adaptation• Multiply each channel by a gain factor
• A more general transformation would correspond to an arbitrary 3x3
matrix
• Best way: gray card• Take a picture of a neutral object (white or gray)
• Deduce the weight of each channel
– If the object is recoded as rw, gw, bw
use weights 1/rw, 1/gw, 1/bw
Slide: F. Durand
![Page 46: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/46.jpg)
White balance
• Without gray cards: we need to “guess” which pixels correspond to white objects
• Gray world assumption• The image average rave, gave, bave is gray
• Use weights 1/rave, 1/gave, 1/bave
• Brightest pixel assumption (non-staurated)• Highlights usually have the color of the light source
• Use weights inversely proportional to the values of the brightest pixels
• Gamut mapping• Gamut: convex hull of all pixel colors in an image
• Find the transformation that matches the gamut of the image to the gamut of a “typical” image under white light
• Use image statistics, learning techniques
Slide: F. Durand
![Page 47: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/47.jpg)
Uses of color in computer vision
Color histograms for indexing and retrieval
Swain and Ballard, Color Indexing, IJCV 1991.
![Page 48: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/48.jpg)
Uses of color in computer vision
Skin detection
M. Jones and J. Rehg, Statistical Color Models with
Application to Skin Detection, IJCV 2002.Source: S. Lazebnik
![Page 49: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/49.jpg)
Uses of color in computer vision
Nude people detection
Forsyth, D.A. and Fleck, M. M., ``Automatic Detection of Human
Nudes,'' International Journal of Computer Vision , 32 , 1, 63-77,
August, 1999
![Page 50: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/50.jpg)
Uses of color in computer vision
Image segmentation and retrieval
C. Carson, S. Belongie, H. Greenspan, and Ji. Malik, Blobworld:
Image segmentation using Expectation-Maximization and its
application to image querying, ICVIS 1999. Source: S. Lazebnik
![Page 51: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/51.jpg)
Uses of color in computer vision
Robot soccer
M. Sridharan and P. Stone, Towards Eliminating Manual
Color Calibration at RoboCup. RoboCup-2005: Robot
Soccer World Cup IX, Springer Verlag, 2006
Source: K. Grauman
![Page 52: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/52.jpg)
Uses of color in computer vision
Building appearance models for tracking
D. Ramanan, D. Forsyth, and A. Zisserman. Tracking People by Learning their
Appearance. PAMI 2007.Source: S. Lazebnik
![Page 53: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/53.jpg)
Interlude
•
•
–
•
![Page 54: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/54.jpg)
Image Filtering
![Page 55: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/55.jpg)
Overview of Filtering
•
•
•
![Page 56: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/56.jpg)
Overview of Filtering
•
•
•
![Page 57: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/57.jpg)
Motivation: Noise reduction
•
Take lots of images and average them!
What’s the next best thing?
Source: S. Seitz
![Page 58: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/58.jpg)
Moving average
111
111
111
“box filter”
Source: D. Lowe
![Page 59: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/59.jpg)
Defining Convolution
lk
lkglnkmfnmgf,
],[],[],)[(
f
Source: F. Durand
• Convention: kernel is “flipped”• MATLAB: conv2 (also imfilter)
![Page 60: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/60.jpg)
Key properties
![Page 61: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/61.jpg)
Properties in more detail
–
–
–
![Page 62: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/62.jpg)
Annoying details
•
–
–
–
f
gg
gg
f
gg
gg
f
gg
gg
full same valid
![Page 63: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/63.jpg)
Annoying details
•
–
–
–
•
•
•
•
Source: S. Marschner
![Page 64: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/64.jpg)
Annoying details
•
–
–
–
•
•
•
•
Source: S. Marschner
![Page 65: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/65.jpg)
Practice with linear filters
000
010
000
Original
?
Source: D. Lowe
![Page 66: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/66.jpg)
Practice with linear filters
000
010
000
Original Filtered
(no change)
Source: D. Lowe
![Page 67: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/67.jpg)
Practice with linear filters
000
100
000
Original
?
Source: D. Lowe
![Page 68: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/68.jpg)
Practice with linear filters
000
100
000
Original Shifted left
By 1 pixel
Source: D. Lowe
![Page 69: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/69.jpg)
Practice with linear filters
Original
?111
111
111
Source: D. Lowe
![Page 70: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/70.jpg)
Practice with linear filters
Original
111
111
111
Blur (with a
box filter)
Source: D. Lowe
![Page 71: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/71.jpg)
Practice with linear filters
Original
111
111
111
000
020
000
- ?
(Note that filter sums to 1)
Source: D. Lowe
![Page 72: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/72.jpg)
Practice with linear filters
Original
111
111
111
000
020
000
-
Sharpening filter- Accentuates differences with local average
Source: D. Lowe
![Page 73: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/73.jpg)
Sharpening
before after
Slide credit: Bill Freeman
![Page 74: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/74.jpg)
Spatial resolution and color
R
G
B
original
Slide credit: Bill Freeman
![Page 75: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/75.jpg)
Blurring the G component
R
G
B
original processed
Slide credit: Bill Freeman
![Page 76: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/76.jpg)
Blurring the R component
original processed
R
G
B
Slide credit: Bill Freeman
![Page 77: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/77.jpg)
Blurring the B component
original
R
G
Bprocessed
Slide credit: Bill Freeman
![Page 78: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/78.jpg)
From W. E.
Glenn, in
Digital
Images and
Human
Vision, MIT
Press,
edited by
Watson,
1993
Slide credit: Bill Freeman
![Page 79: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/79.jpg)
Lab color components
L
a
b
A rotation of the
color
coordinates into
directions that
are more
perceptually
meaningful:
L: luminance,
a: red-green,
b: blue-yellow
Slide credit: Bill Freeman
![Page 80: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/80.jpg)
Blurring the L Lab component
L
a
b
original processed
Slide credit: Bill Freeman
![Page 81: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/81.jpg)
original
Blurring the a Lab component
L
a
b
processed
Slide credit: Bill Freeman
![Page 82: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/82.jpg)
Blurring the b Lab component
original
L
a
bprocessed
Slide credit: Bill Freeman
![Page 83: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/83.jpg)
Overview of Filtering
•
•
•
![Page 84: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/84.jpg)
Smoothing with box filter revisited
• Smoothing with an average actually doesn‟t compare
at all well with a defocused lens
• Most obvious difference is that a single point of light
viewed in a defocused lens looks like a fuzzy blob; but
the averaging process would give a little square
Source: D. Forsyth
![Page 85: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/85.jpg)
Smoothing with box filter revisited
• Smoothing with an average actually doesn‟t compare
at all well with a defocused lens
• Most obvious difference is that a single point of light
viewed in a defocused lens looks like a fuzzy blob; but
the averaging process would give a little square
• Better idea: to eliminate edge effects, weight
contribution of neighborhood pixels according to their
closeness to the center, like so:
“fuzzy blob”Source: D. Forsyth
![Page 86: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/86.jpg)
Gaussian Kernel
• Constant factor at front makes volume sum to 1 (can be
ignored, as we should re-normalize weights to sum to 1 in any case)
0.003 0.013 0.022 0.013 0.0030.013 0.059 0.097 0.059 0.0130.022 0.097 0.159 0.097 0.0220.013 0.059 0.097 0.059 0.0130.003 0.013 0.022 0.013 0.003
5 x 5, = 1
Source: C. Rasmussen
![Page 87: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/87.jpg)
Choosing kernel width
• Gaussian filters have infinite support, but
discrete filters use finite kernels
Source: K. Grauman
![Page 88: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/88.jpg)
Choosing kernel width
• Rule of thumb: set filter half-width to about
3 σ
![Page 89: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/89.jpg)
Example: Smoothing with a Gaussian
![Page 90: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/90.jpg)
Mean vs. Gaussian filtering
![Page 91: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/91.jpg)
Gaussian filters
• Remove “high-frequency” components from
the image (low-pass filter)
• Convolution with self is another Gaussian• So can smooth with small-width kernel, repeat, and get
same result as larger-width kernel would have
• Convolving two times with Gaussian kernel of width σ is
same as convolving once with kernel of width σ√2
• Separable kernel• Factors into product of two 1D Gaussians
Source: K. Grauman
![Page 92: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/92.jpg)
Separability of the Gaussian filter
Source: D. Lowe
![Page 93: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/93.jpg)
Separability example
*
*
=
=
2D convolution
(center location only)
Source: K. Grauman
The filter factors
into a product of 1D
filters:
Perform convolution
along rows:
Followed by convolution
along the remaining column:
For MN image, PQ filter: 2D takes MNPQ add/times,
while 1D takes MN(P + Q)
![Page 94: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/94.jpg)
Overview of Filtering
•
•
•
![Page 95: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/95.jpg)
Alternative idea: Median filtering
• A median filter operates over a window by
selecting the median intensity in the window
• Is median filtering linear?Source: K. Grauman
![Page 96: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/96.jpg)
Median filter
Replace each pixel by the median over N pixels (5 pixels, for these examples). Generalizes to “rank order” filters.
5-pixel neighborhood
In: Out:
In: Out:
Spike noise is removed
Monotonic edges remain unchanged
Median([1 7 1 5 1]) = 1Mean([1 7 1 5 1]) = 2.8
![Page 97: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/97.jpg)
Median filtering results
http://homepages.inf.ed.ac.uk/rbf/HIPR2/mean.htm#guidelines
![Page 98: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/98.jpg)
Median vs. Gaussian filtering3x3 5x5 7x7
Gaussian
Median
![Page 99: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/99.jpg)
Edges
http://todayinart.com/files/2009/12/500x388xblind-contour-line-drawing.png.pagespeed.ic.DOli66Ckz1.png
![Page 100: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/100.jpg)
Edge detection
• Goal: Identify sudden
changes (discontinuities) in
an image• Intuitively, most semantic and shape
information from the image can be
encoded in the edges
• More compact than pixels
• Ideal: artist‟s line drawing
(but artist is also using
object-level knowledge)
Source: D. Lowe
![Page 101: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/101.jpg)
Origin of edges
Edges are caused by a variety of factors:
depth discontinuity
surface color discontinuity
illumination discontinuity
surface normal discontinuity
Source: Steve Seitz
![Page 102: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/102.jpg)
Edges in the Visual Cortex
Extract compact, generic, representation of image
that carries sufficient information for higher-level
processing tasks
Essentially what area
V1 does in our visual
cortex.
http://www.usc.edu/programs/vpl/private/photos/research/retinal_circuits/figure_2.jpg
![Page 103: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/103.jpg)
The gradient points in the direction of most rapid increase
in intensity
Image gradient
The gradient of an image:
The gradient direction is given by
Source: Steve Seitz
The edge strength is given by the gradient magnitude
• How does this direction relate to the direction of the edge?
![Page 104: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/104.jpg)
Differentiation and convolution
Recall, for 2D function,
f(x,y):
This is linear and shift
invariant, so must be
the result of a
convolution.
We could approximate
this as
(which is obviously a
convolution)
f
x lim
0
f x , y
f x,y
f
xf xn1,y f xn , y
x
-1 1
Source: D. Forsyth, D. Lowe
![Page 105: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/105.jpg)
Finite difference filters
Other approximations of derivative filters exist:
Source: K. Grauman
![Page 106: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/106.jpg)
Finite differences: example
Which one is the gradient in the x-direction (resp. y-direction)?
![Page 107: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/107.jpg)
Effects of noise
Consider a single row or column of the image• Plotting intensity as a function of position gives a signal
Where is the edge?Source: S. Seitz
![Page 108: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/108.jpg)
Effects of noise
• Finite difference filters respond strongly to
noise• Image noise results in pixels that look very different from
their neighbors
• Generally, the larger the noise the stronger the response
• What is to be done?• Smoothing the image should help, by forcing pixels different
from their neighbors (=noise pixels?) to look more like
neighbors
Source: D. Forsyth
![Page 109: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/109.jpg)
Solution: smooth first
• To find edges, look for peaks in )( gfdx
d
f
g
f * g
)( gfdx
d
Source: S. Seitz
![Page 110: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/110.jpg)
• Differentiation is convolution, and convolution
is associative:
• This saves us one operation:
gdx
dfgf
dx
d )(
Derivative theorem of convolution
gdx
df
f
gdx
d
Source: S. Seitz
![Page 111: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/111.jpg)
Derivative of Gaussian filter
Which one finds horizontal/vertical edges?
x-direction y-direction
![Page 112: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/112.jpg)
Smoothed derivative removes noise, but blurs
edge. Also finds edges at different “scales”.
1 pixel 3 pixels 7 pixels
Scale of Gaussian derivative filter
Source: D. Forsyth
![Page 113: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/113.jpg)
• The gradient magnitude is large along a thick
“trail” or “ridge,” so how do we identify the actual
edge points?
• How do we link the edge points to form curves?
Implementation issues
Source: D. Forsyth
![Page 114: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/114.jpg)
Designing an edge detector
• Criteria for an “optimal” edge detector:• Good detection: the optimal detector must minimize the
probability of false positives (detecting spurious edges caused by
noise), as well as that of false negatives (missing real edges)
• Good localization: the edges detected must be as close as
possible to the true edges
• Single response: the detector must return one point only for each
true edge point; that is, minimize the number of local maxima
around the true edge
Source: L. Fei-Fei
![Page 115: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/115.jpg)
Canny edge detector
• This is probably the most widely used edge
detector in computer vision
• Theoretical model: step-edges corrupted by
additive Gaussian noise
• Canny has shown that the first derivative of
the Gaussian closely approximates the
operator that optimizes the product of signal-
to-noise ratio and localization
• MATLAB: edge(image, „canny‟)
J. Canny, A Computational Approach To Edge Detection, IEEE
Trans. Pattern Analysis and Machine Intelligence, 8:679-714, 1986.
Source: L. Fei-Fei
![Page 116: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/116.jpg)
Canny edge detector
1. Filter image with derivative of Gaussian
2. Find magnitude and orientation of gradient
3. Non-maximum suppression:• Thin multi-pixel wide “ridges” down to single pixel width
Source: D. Lowe, L. Fei-Fei
![Page 117: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/117.jpg)
Non-maximum suppression
At q, we have a
maximum if the
value is larger
than those at
both p and at r.
Interpolate to
get these
values.
Source: D. Forsyth
![Page 118: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/118.jpg)
Example
original image (Lena)
![Page 119: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/119.jpg)
Example
norm of the gradient
![Page 120: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/120.jpg)
Example
thresholding
![Page 121: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/121.jpg)
Example
Non-maximum suppression
![Page 122: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/122.jpg)
Canny edge detector
1. Filter image with derivative of Gaussian
2. Find magnitude and orientation of gradient
3. Non-maximum suppression• Thin multi-pixel wide “ridges” down to single pixel width
4. Linking of edge points
Source: D. Lowe, L. Fei-Fei
![Page 123: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/123.jpg)
Assume the marked
point is an edge point.
Then we construct the
tangent to the edge
curve (which is normal
to the gradient at that
point) and use this to
predict the next points
(here either r or s).
Edge linking
Source: D. Forsyth
![Page 124: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/124.jpg)
Canny edge detector
1. Filter image with derivative of Gaussian
2. Find magnitude and orientation of gradient
3. Non-maximum suppression• Thin multi-pixel wide “ridges” down to single pixel width
4. Linking of edge points• Hysteresis thresholding: use a higher threshold to start
edge curves and a lower threshold to continue them
Source: D. Lowe, L. Fei-Fei
![Page 125: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/125.jpg)
Hysteresis thresholding
• Use a high threshold to start edge curves and
a low threshold to continue them• Reduces drop-outs
Source: S. Seitz
![Page 126: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/126.jpg)
Hysteresis thresholding
original image
high threshold
(strong edges)
low threshold
(weak edges)
hysteresis threshold
Source: L. Fei-Fei
![Page 127: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/127.jpg)
Effect of (Gaussian kernel spread/size)
Canny with Canny with original
The choice of depends on desired behavior• large detects large scale edges
• small detects fine features
Source: S. Seitz
![Page 128: Lecture 2: Color, Filtering & Edgesfergus/teaching/vision_2012/2_Color_Filtering... · Color matching experiment 2 p p 1 p 2 p 3 1 p 2 p 3 We say a “negative” amount of p 2 was](https://reader033.vdocument.in/reader033/viewer/2022042805/5f5ebf6055fee0314d66cb8f/html5/thumbnails/128.jpg)
Edge detection is just the beginning…
Berkeley segmentation database:http://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/segbench/
image human segmentation gradient magnitude