Transcript Color and Image Manipulation

Color
CSC361/661 -- Digital Media
Spring 2002
Burg/Wong
1
Color


Light is electromagnetic energy in the 400- to
700 nanometer wavelength part of the
spectrum, perceived as the colors ranging
through violet, indigo, blue, green, yellow,
orange, and red.
The colors that we see in the world around us
are generally not pure colors consisting of a
single wavelength. Rather, color sensation
results from the dominant wavelength of the
light reflecting off or emanating from an
object.
2
Color Terminology






Hue – color
Monochromatic color – a color that is created from only one
wavelength. (Most colors that we experience are NOT
monochromatic. They result from a combination of wavelengths.
The dominant wavelength gives us our color sensation.)
Chrominance – color information
Luminance – lightness or brightness information
Additive color systems – based on adding colored light (as in
computer monitors). A combination of all colors gives white.
Subtractive color systems – based on adding pigments (as in
printing). A combination of all colors gives black.
3
Human Perception of Color



To humans, color sensation is a matter of
subjective perception resulting from the effect
of light on the cones of the eyes.
There are three types of cones, each one
with a particular sensitivity to red, green, or
blue light.
This decomposition of light into three color
components is called the tristimulus theory of
color and is the basis for the RGB color
model.
4
Response of Cones in the Eyes

This graphs shows the results
of experiments in which the
fraction of light absorbed by
each type of cone is measured
for each color in the visible
spectrum. The green cone
absorbs the most light. (Note
that this is essentially the same
graph as Figure 2.1 of the
handout given in class, except
that the units are different,
eliminating the negative
values.)
5
The Eye’s Sensitivity to
Different Colors

This graph shows the eye’s
overall response as the dominant
wavelength (i.e., the hue) is
varied across the visible
spectrum. (Luminance is kept
constant.) The eye’s sensitivity
peaks at around 550 nm, the
wavelength of yellow-green light.
Note that this graph is the sum
of the three graphs in the
previous slide.
6
Testing Color Response for
RGB Model
To create the RGB color model, it is necessary to
determine how much of each color component is
needed to create all the dominant wavelengths in
the visible spectrum.
 This is done by projecting “pure” colors onto one
screen, mixing amounts of R, G, and B on a
neighboring screen, and asking a large number of
people to say when the colors match.
 Matching color C is expressed by
C = R*R + G*G + B*B

7
RGB Color Matching

This graph shows the
amounts of red, green, and
blue light needed by an
average observer to match
color samples as they vary
across the spectrum.
(Luminance is kept
constant.) A negative value
means it is not possible to
match the original color with
RGB as primaries, so some
R, G, or B has to be added
to the original color sample.
8
RGB Color on a Computer


The RGB color model works well for
computers because it matches the technology
of monitors.
On a color monitor, color is produced by
exciting three adjacent dots made of red,
green, and blue phosphors. Because the dots
are so small, they are blended into one color
by the eye. Note that the color is not
blended by putting one color of light over
another – it is blended by the eye.
9
RGB Color on a Computer



Not all colors perceivable by humans can be
shown on a computer screen with the RGB model.
The same RGB values will not necessarily result in
the same colors on two different monitors
because monitors are not calibrated to a single
standard.
RGB colors are not pure, saturated colors. This is
because the kind of light emitted by an excited
phosphor is not of a single wavelength, but has a
spectral power distribution over a band of
frequencies.
10
RGB Color on a Computer

RGB is perceptually non-linear. This means
that equal differences in the RGB values do
not correspond to equal differences in the
perceived color. Low RGB values produce
small changes in color on the screen (as you
move from one low value to the next). Large
RGB values produce very perceivable
differences as you move from one high RGB
value to the next.
11
CIE Color


The graph on slide 13 shows that not all visible
colors can be represented with RGB as primaries.
(You can’t add positive amounts of RGB to get all
the colors. In some cases, you have to “take some
color away” from the original sample being
matched.)
The Commission Internationale de l’Eclairage (CIE)
decided that we need a standard color model that
is based on primaries which, when mixed together,
produce all the visible colors.
12
CIE Standardization

Another motivation for the CIE
model is that the RGB and
CMYK models are device
dependent. That is different
monitors use different R, G,
and B colors of phosphors.
Different printers use different
CMYK colors of ink. CIE
standardization gives us a way
to map between systems.
13
CIE Color Primaries


The CIE color primaries X, Y, and Z
replace R, G, and B. X, Y, and Z are
“artificial primaries,” not visible colors
like R, G, and B.
These primaries can be combined in
various proportions to produce all the
colors the human eye can see.
14
CIE Color Matching

This graph shows the
amounts of X, Y, and Z light
needed by an average
observer to match color
samples as they vary across
the spectrum. (Luminance
is kept constant.) Notice
that no negative values are
needed.
15
CIE Color Space




The graph on the previous slide
shows how X, Y, and Z can be
combined to create any visible
color, where all the colors in the
spectrum are considered at the
same luminance.
This graph shows the entire CIE
color space, where not only the
color but the luminance varies.
In the CIE color model,
a color C is given by
C = X*X + Y*Y + Z*Z
16
CIE Color Model on the
X+Y+Z = 1 Plane

If we want to consider each component
as a percentage of the total amount of
light, we can “normalize” the values:
X
x
X Y  Z
Y
y
X Y  Z
Z
z
X Y  Z
Note: X + Y + Z is the total
amount of light energy.
Also note that x + y + z = 1
17
CIE Color Space


This graph shows the amounts of
X, Y, and Z needed for all colors
in the visible spectrum. The
X + Y + Z = 1 plane is shown as
the triangle embedded in the
graph.
The x, y, and z computed on the
last slide lie on the X+Y+Z=1
plane. It is convenient to
consider this plane only, which
effectively reduces our
consideration to constant
luminance.
18
CIE Chromaticity Diagram



Picture the portion of
the CIE color space that
intersects the X+Y+Z=1
plane.
Now picture projecting
that part of the
X+Y+Z=1 plane down
onto the X, Y plane.
This is how the CIE
Chromaticity Diagram is
created (next slide).
19
CIE Chromaticity Diagram



This graph represents the hue
and saturation of all colors in the
visible spectrum. This is
“chromaticity” information.
All perceivable colors with the
same chromaticity but differet
luminances map into the same
point in this graph.
The 100 percent spectrally pure
colors are on the curved
perimeter of the graph. The dot
in the center represents the
chromaticity of daylight (white
light).
20
Dominant Wavelength on CIE
Color Diagram


To determine what the
dominant wavelength of a
color A is from the diagram,
draw a line between C (white)
and the closest perimeter. The
dominant wavelength is at B.
The degree of saturation is
given by the proportion of
segment AB to segment CB.
The closer A is to the
perimeter, the more saturated
the color.
21
Color Gamuts Represented on
CIE Diagram

All colors on the line
IJ can be created by
additively mixing
colors I and J; all
colors in the triangle
IJK can be created
by mixing colors I, J,
and K.
22
RGB Color Gamut in Terms of
CIE Diagram




We can see how much of the
visible spectrum is displayable
on a computer monitor by
looking at the gamut within
the RGB diagram, represented
by the triangle.
Phosphors on a computer
monitor have these
approximate values:
red
green blue
x ~.61 ~.25 ~.15
y ~.34 ~.62 ~.063
23
CMYK model



CMYK is primarily a printing color
model.
Cyan, magenta, and yellow are called
the subtractive primaries.
In practice, cyan, magenta, and yellow
don’t produce all the colors needed for
printing. Blacks come out muddy. So a
pure black is added in. That’s the K.
24
CMYK Model




Cyan, magenta, yellow, and
black
Cyan is white light with red
taken out.
C=G+B=W-R
Magenta is white light with
green taken out.
M=R+B=W-G
Yellow is white light with
blue taken out.
Y=R+G=W-B
25
CMYK vs. RGB

The colors printable with
the CMYK model do not
overlap exactly with the
colors displayable on an
RGB monitor, as
represented by their
respective gamuts within
the CIE diagram.
26
Hue, Saturation, and Lightness





One way to represent color is by dividing it into its
hue, saturation, and lightness components.
Hue (or color) is determined by the dominant
wavelength.
Saturation is a matter of how much white light is
added in. The less white light, the more saturated
the color.
Lightness is how much black is in the color.
Hue and saturation are elements of chrominance.
Lightness is a matter of luminance.
27
HSV


HSV stands for hue,
saturation, and value (where
value represents lightness or
brightness).
Some people find the HSV
model more intuitive than
RGB. It is easier to think of
colors in terms of their hue,
tint, and shade rather than as
combinations of red, green,
and blue.
28
YUV Color Model


YUV is a general term that refers to any
color model that has one luminance
component (Y) and two chrominance (i.e.,
color) components (U and V). (You’ll also
see references to the YIQ model, which is
the same thing.)
Y’CBCR is a specific instance of a YUV
model.
29
YUV Color Model


YUV is a color model appropriate to color TV
because it makes it possible to send the color
information separate from the luminance
information, so that signals for black and
white vs. color TV are easily separated.
YUV is also a good representation for
compression, because some of the
chrominance information can be thrown out
without loss of quality in the picture (since
the human eye is less sensitive to
chrominance than luminance).
30