12
Displays for Embedded Systems and
Physiological Limits on Perceptible Outputs
Figure 12.1: Spot the
differences.
In this discussion the names of colors — “red”, “green”, “blue” and so
on — will be reserved for the color sensation we have when we look at
the world around us. In short, only our eyes can categorize the color of
objects; spectrophotometers cannot. This point is not a trivial one
because many people viewing some of our experiments for the first
time will identify something as being red or green but will then ask, as
if their eyes were being fooled. “What color is it really?” The answer is
that the eye is not being fooled. It is functioning exactly as it must
with involuntary reliability to see constant colors in a world
illuminated by shifting and unpredictable fluxes of radiant energy.
— Edwin H. Land, “The Retinex Theory of Color Vision”, Scientific
American, December 1977, Vol. 237, No. 6, pp. 108–128.
Table 12.1: Concepts.
Concept
Human color perception § 12.2.
Image quality metrics §
12.3.
Display technology §
12.4.
Displays and perception §
12.5.
End-to-end systems §
12.6.
Consider a computing system that performs a task whose end result is
only intended for display to a human observer. If the task can be altered
such that it uses fewer resources, and if most human observers (or a specific
one) cannot perceive any change in the end result, then the system can be
made more efficient by either avoiding work or making errors as long as
2 phillip stanley-marbell
these errors are not perceptible.
For computing tasks such as numeric solution of differential equations,
computations which are part of a program for completing a tax return, or
computing the cost of a sales order, obtaining the correct numeric result is
almost always critical. Similarly, an alphanumeric display such as that dis-
playing departures and arrivals at an airport or train station must display the
exact information sent to it in order to be useful.
But not all computations and not all displayed information is on data that
have a precise or quantitative nature. Computations whose results only feed
into determining the color of a pixel in a temporary on-screen image have
their requirements on accuracy bounded by the limits of human color percep-
tion (and attention spans). Similarly, because the display panel on a phone
or smart watch can consume different amounts of power based only on the
color content of images (Figure
12.2), changes to the color and shape content
of images can affect the power dissipation of the display.
()
(a)
(b)
Figure 12.2: The power
dissipation for a repre-
sentative OLED panel as
a function of a range of
fully-saturated hues (the
hue space wraps around),
shown in linear coordinates
(a) and in polar coordinates
(b).
Because displays constitute a large fraction of the power dissipation of
many modern mobile platforms it is possible to improve the battery life of
many platforms if images could be adapted in ways that exploit display prop-
erties without being visually perceptible. In order to perform such changes
however, we would need to have quantitative answers to several questions,
such as:
• How sensitive are users to changes in color?
• Are there colors that are indistinguishable to humans but lead to signifi-
cant changes in display power dissipation of some display types?
• How sensitive are users to changes in shape?
12.1 Intended Learning Outcomes
At the end of this chapter, you should be able to:
• Explain the difference between spectral colors and perceived colors.
foundations of embedded systems 3
• Recall and explain the existing theories for how the wavelengths in a beam
of light lead to perceived colors in humans.
• Enumerate the basic properties of human color vision that have a bearing
on the design of displays for embedded systems
• Enumerate the properties of traditional liquid crystal displays (LCDs) and
organic light-emitting diode (OLED) displays.
• Propose and design changes to existing system designs that use OLED
displays in order to improve their energy efficiency.
12.1.1 Learning outcomes pre-assessment
Complete the following
quiz to evaluate your prior knowledge of the material
for this chapter.
12.1.2 Things to think about
Complete the following
thinking exercise to stimulate your thoughts about
the contents of this chapter before proceeding with the material.
12.1.3 Things to look out for
Concepts people sometimes get confused by in this chapter include:
1. Why long-, medium- and short-wavelength cones are not present in equal
proportion in the human retina.
2. Why human color vision is more sensitive in the “green” part of the vis-
ible spectrum when we have more long-wavelength (“red”) cones than
medium wavelength (“green”) cones.
3. The concept of hue.
4. How wavelengths translate to the hue scale.
5. How the Land experiments
1
led to perceived full-color images when il-
1
Land
1983; Land 1986;
Land
1977; Land 1959c;
Land
1959b; Land 1959a.
luminated with monochromatic light from only the “yellow” part of the
visible spectrum.
6. How sub-pixel arrangements in displays affect color perception.
7. What/how many parameters define color.
4 phillip stanley-marbell
12.1.4 The muddiest point
As you go through the material in this chapter, think about the following two
questions and note your responses for yourself or using the annotation tools
of the online version of the chapter. You will have the opportunity to submit
your responses to these questions at the end of the chapter:
1. What is least clear to you in this chapter? (You can simply list the section
numbers or write a few words.)
2. What is most clear to you in this chapter? (You can simply list the section
numbers or write a few words.)
12.2 Human Perception of Color
Humans with normal color vision have three types of color sensitive cells
(cones) in their retinas. Each of the three types of cone cells is sensitive
to a broad range of wavelengths of light, but the different types each have
peak sensitivities in the short-, medium- and long-wavelength portions of
the visible spectrum. Because of the locations of these peak sensitivities, most
people have most of their color sensitivity in the green portion of the visible
spectrum. However, even though most of the sensitivity is to the portion of
the visible spectrum close to green, most people are not necessarily able to
easily distinguish between different wavelengths of light in the green portion
of the visible spectrum as well as they do for other wavelengths.
For beams of light made up of a single wavelength (spectral colors), con-
trolled colorimetric studies have been used to quantify how much a spec-
tral color of a given wavelength must be changed to longer or shorter wave-
lengths for the change to be perceptible by humans. Several of these studies,
such as studies by Wright and Pitt in 1934, and by Bedford and Wyszecki
in 1958
2
showed that the ability of human observers to differentiate spec-
2
Wyszecki and Stiles
2000.
tral colors varies with wavelength. Changes in wavelength of spectral colors
at wavelengths close to 450 nm (approximately, indigo) and 525 nm (approx-
imately, green) were markedly more difficult for observers to differentiate
than changes in other parts of the visible spectrum outside of the extremes.
At 450 nm and 525 nm, wavelength changes of up to 4 nm were necessary for
an observer to perceive a change in color, compared to wavelength changes
of just 1 nm which were perceptible in the rest of the central portion of the
visible spectrum.
The perception of spectral colors in color-matching experiments is differ-
ent from the perceived color in images. Contrary to common belief, the col-
ors humans perceive in natural images, in contrast to the color sensations
perceived in color-matching experiments, are not a direct result of the wave-
lengths of spectral light or combinations of the spectra that objects reflect or
foundations of embedded systems 5
project. In a series of experiments in 1959
3
and developed over twenty years,
3
Land
1983; Land 1986;
Land
1977; Land 1959c;
Land
1959b; Land 1959a.
Edwin Land (the founder of the Polaroid Corporation) and his colleagues first
showed how pairs of bands of wavelengths (e.g., red and white) when used
to illuminate a pair of black-and-white negatives taken with red and green
filters yield full color perceived images, even though the same illumination
without the black-and-white negatives simply yielded a pink screen. Land
and his colleagues also later demonstrated that when they illuminated differ-
ently color patches with controlled intensities of red, green, and blue light, so
that the amount of red, green, and blue reflected off the patches and reaching
the eye were identical, observers would still observe different colors. These
and other investigations led them to the conclusion that perceived color was
not just the result of the magnitudes of the constituent wavelengths of light
reaching the observer, but rather the result of the relative spatial distribu-
tions: The visual system captures images corresponding to short- medium-
and long-wavelength photoreceptors in the retina (the s-, m-, and l- cones)
and does not simply average these images. Instead, it compares the relative
intensities of nearby points in the scene captured by the normal human eye’s
three photoreceptors, to deduce what we perceive as color.
12.3 Quantifying Errors and Efficiency: SSE, MSE, SNR, PSNR,
and SSIM
A way to quantify errors and efficiency is a prerequisite to meaningfully
trading errors for efficiency. For images, there are several commonly-used
ways to quantify errors.
Let r be a reference image and let t be an image to be compared to r, with
both r and t being w by h pixels in size. The most basic measure of difference
between the original and modified images is the sum of squared errors (SSE)
between the reference image r and the transformed image t, defined as
SSE =
w− 1
∑
x=0
h− 1
∑
y=0
r
x,y
− t
x,y
2
(12.1)
The mean squared error (MSE) is defined as
MSE =
1
w · h
SSE. (12.2)
The signal to noise ratio (SNR) is the ratio of the sum of squared signal to the
SSE
SNR = 10 log
10
"
∑
w− 1
x=0
∑
h− 1
y=0
r
2
x,y
SSE
#
(12.3)
6 phillip stanley-marbell
The peak signal to noise ratio (PSNR) takes the ratio of the maximum of all
the squared signal (instead of the sum) to the mean squared error (instead of
the SSE):
PSNR = 10 log
10
"
max(r
2
x,y
)
1
w· h
SSE
#
(12.4)
Let µ
r
and µ
t
be the mean pixel values for two grayscale images r and t. Let
σ
2
rt
be the covariance between the images r and t and let σ
r
and σ
t
be the
standard deviations of pixel values in r and t. Let C
1
and C
2
be constants
derived from the dynamic range of the image representation format with b
bits per pixel and equal to (0.01 · 2
b
− 1)
2
and (0.03 · 2
b
− 1)
2
respectively.
Then, the structural similarity (SSIM) between the two images r and t is
defined as:
SSIM(r, t) =
(2µ
r
µ
t
+ C
1
)(2σ
rt
+ C
2
)
(µ
2
r
+ µ
2
t
+ C
1
)(σ
2
r
+ σ
2
t
+ C
2
)
(12.5)
12.4 Display Technology
Displays account for up to 40% of the power usage in mobile devices such
as phones, smart watches, and tablets. This makes it of great interest to
manufacturers to develop ways to reduce display energy usage, as doing so
would enable improvements in battery lifetime.
The most common type of display is the liquid crystal display or LCD.
LCDs consist of a light source, called a backlight, which illuminates a col-
lection of color filters at the individual pixels; whether each of these color
filters receives light is controlled by a layer beneath the color filters, made
up of (liquid) crystals, which can be made opaque or transparent, acting like
miniature shutters, based on an electrical signal. Because the backlight is on
all the time even though individual pixels might be blocking its light, the
power dissipation of LCD displays is dominated by the backlight.
In contrast to LCDs, which have a single light source, organic light-emitting
diode (OLED) displays typically have dedicated red, green, and blue light
sources made from organic compounds, for each individual pixel. Because
they are made up of a smaller number of layers (no need for the LCD layer
or for a color filter), OLED displays can often be made thinner than LCDs.
The efficiencies of converting electrical current into light in red, green, and
blue OLED pixels varies across the different organic compounds employed in
making them. For example, on some OLED displays, a fully-bright blue pixel
can dissipate almost twice as much power as a fully-bright green one. And
over the lifetime of display, the efficiencies also degrade, with blue sub-pixels
degrading in efficiency more quickly than red and green pixels. As a result,
to achieve the same brightness as an OLED panel ages, blue pixels need to
foundations of embedded systems 7
be driven with higher currents (and hence dissipate more power) to achieve
the same level of brightness as blue pixels on a new OLED display. This
challenge is referred to by researchers in the area of organic LEDs as “the
blue problem”. Because of these varied phenomena, the amount of power
dissipated by an OLED display depends on the image being displayed.
Almost a decade ago, a number of researchers, starting with Lin Zhong
at Rice University in Houston Texas, begun exploiting the color-dependence
of OLED power dissipation to improve the efficiency of OLED displays by
changing colors in ways in which an observer might find acceptable (such as
by changing an operating system’s or individual application’s color theme),
but for which the change leads to a reduction in display power usage.
12.5 Exploiting Perception for Display Energy Efficiency
The interfaces for surfacing perceptual signals, such as displays and audio,
contribute an increasing fraction of system energy usage in wearable and
mobile systems. Because the phenomena underlying their operation (e.g.,
photon generation, mechanical displacement) are less amenable to improve-
ments in transistor properties than computation is, their relative importance
will likely grow in the future.
A number of techniques for error efficiency, targeted primarily at legacy
backlit LCDs, have been developed to reduce display power dissipation.
4
4
Cheng, Hou, and Pedram
2004; Ranganathan et al.
2006.
Prior work on trading image fidelity for power or performance can be clas-
sified broadly into six directions: Color transformation by convex optimiza-
tion; color transformation in restricted applications such as web browsers and
by color remapping; color transformation by electrical control of the display
panel; selective dimming based on a user’s visual focus; and image fidelity
tradeoff analyses that employ perceptual user studies.
12.5.1 Color transformation by convex optimization
Reducing display power dissipation under an image-distortion constraint can
be framed as an optimization problem. How efficiently one can solve this op-
timization problem, and the quality of the solution (global versus local min-
ima) however depends on the power model and perceptual model that are
used in the optimization problem formulation. Dong et al. employ a power
model that has a count of parameters that is exponential in the display’s color
depth (e.g., 3 · 2
3·8
or 48 million parameters for a 24-bit color display).
5
Their
5
Dong and Zhong
2011.
color optimization is thus expensive to do even offline, and not practical to
do online (i.e., in real-time). Because the optimal solution of their optimiza-
tion problem requires an operation count that is exponential in the number
of pixels, the authors propose a greedy heuristic that is O(n
3
) for n-pixel im-
8 phillip stanley-marbell
ages. As a result of their formulation, their color optimizations take many
hours to complete.
6
6
Dong and Zhong
2011.
12.5.2 Application-specific color transformations and color remapping
When color transformations are applied in restricted contexts such as in color
schemes for infographics
7
or in graphical user interface (GUI) color schemes,
8
7
Chuang, Weiskopf, and
Möller
2009; Wang, Lin,
and North
2012.
8
Dong, Choi, and Zhong
2009a; Dong, Choi, and
Zhong
2009b; Dong and
Zhong
2011.
colors that are more power-expensive on OLED displays may be substituted
for ones that lead to lower display power dissipation.
Another approach to exploiting the color dependence of OLED power dis-
sipation, for improved energy efficiency is to directly modify individual ap-
plications such as games,
9
web browsers,
10
and web servers
11
which provide
9
Anand et al.
2011.
10
Dong and Zhong
2011;
Li, Tran, and Halfond
2015.
11
Li, Tran, and Halfond
2014.
content to devices with OLED displays. Application-specific tradeoff tech-
niques have the disadvantage that modifications must be repeated for each
new application. Application-specific techniques can also be complex: For
example, the color-adaptive Chameleon web browser
12
employs several tech-
12
Dong and Zhong
2011.
niques including designing color schemes for specific popular websites, in-
verting colors in web pages, and requiring users to explicitly select schemes.
Chameleon requires color maps to be calculated offline, using an optimiza-
tion method which the authors themselves describe as “compute-intensive”,
partly because it requires a large number of parameters: 3 · 2
b
for a display
with b bits of color depth.
12.5.3 Power characterization and color transformation by electrical control
One way to control the power dissipation of displays is to manipulate their
electrical interfaces, such as their backlights, display drivers, and so on, so
that these components use less power. To understand the effect that such
electrical control has on image quality however requires detailed electrical
and perceptual characterization. As a result of this need for characterization,
there have been several studies of LCD power dissipation as a function of
backlight level and perceptual metrics,
13
as well as studies of OLED power
13
Chang, Choi, and Shim
2004; Choi, Shim, and
Chang
2002; Cheng, Hou,
and Pedram
2004; Cheng,
Hsu, and Chao
2006;
Schuchhardt et al.
2015.
dissipation as a function of color and luminance,
14
and models to estimate
14
Dong, Choi, and Zhong
2009a; Shin et al. 2011; Mit-
tal, Kansal, and Chandra
2012; Chen et al. 2013.
OLED display power.
15
15
Harter et al.
2004.
Once displays have been characterized, they can also be controlled to ex-
ploit properties observed in the characterization. The power versus percep-
tual quality tradeoff techniques that have been explored for LCDs include
contrast scaling,
16
luminance scaling,
17
and tone mapping.
18
A number of
16
Cheng, Hou, and Pedram
2004; Pasricha et al. 2004.
17
Chang, Choi, and Shim
2004.
18
Iranli and Pedram
2005;
Anand et al.
2011.
recent research efforts have attempted to apply similar hardware techniques
to OLED displays. These techniques have included dynamic voltage scaling
of OLED display driver amplifiers.
19
19
Shin et al.
2011.
foundations of embedded systems 9
12.5.4 Selective area dimming
A number of research efforts selectively dim portions of an OLED display
panel, based on heuristics of a user’s focus of attention.
20
The techniques are
20
Tan et al.
2013.
obtrusive, and when they guess the user’s focus of attention incorrectly, can
render a device unusable. Other research efforts have used heuristics to guess
which part of a display is occluded by a user’s hand;
21
these latter techniques
21
Chen et al.
2014.
must, among other things, guess whether a user is left- or right-handed, how
large their hands are, whether they are using a stylus, and so on.
12.5.5 User studies of image fidelity versus power tradeoffs
The value of techniques that trade perceptual quality for power or perfor-
mance depend on how accurately the techniques quantify their effect on
human perception. The two primary approaches to quantifying perceptual
quality are quantitative metrics such as the structural similarity (SSIM) and
peak signal to noise ration (PSNR), introduced in Section
12.3, and human
user studies.
Several studies of the tradeoffs between image quality and power dissi-
pation of displays have employed perceptual studies. These studies have
however all involved only a small number of participants. For example, Har-
ter et al.
22
employed a study size of 12 users in their analysis of the effects
22
Harter et al.
2004.
of selective display area dimming for OLED displays, while Tan et al.
23
em-
23
Tan et al. 2013.
ployed 30 users in evaluating a similar technique. Li et al.
24
conducted a
24
Li, Tran, and Halfond
2014.
perceptual study with 17 users to evaluate their color-adaptive server-side
color transformations, while Dong et al.
25
employ 20 participants to evaluate
25
Dong and Zhong
2011.
their color-adaptive web browser. Anand et al.
26
conducted a user study with
26
Anand et al. 2011.
60 users to evaluate a display brightness and image tone mapping technique.
All of these prior efforts provided valuable insight into the challenges and
benefits of performing perceptual user studies.
12.6 Exploiting Perceptual Flexibility in End-to-End Systems
Crayon
27
is a system for exploiting perceptual flexibility to reduce display
27
Stanley-Marbell, Estellers,
and Rinard
2016.
power dissipation. At the core of the Crayon system is an intermediate rep-
resentation for representing bitmapped graphics (e.g., photographic images)
and vector graphics (e.g., the drawing operations in a user interface). Across
these representations, Crayon applies several techniques to reduce power dis-
sipation in exchange for visual accuracy.
In bitmap images, a color transform is applied to each individual pixel in-
dependently in order to reduce power dissipation. The transform optimizes
a simple convex function that trades visual fidelity (with a least-squares
penalty on deviation from the target color) for reduced power dissipation; the
10 phillip stanley-marbell
penalty is based on an experimentally-measured model of the power dissipa-
tion for each color. The result of the optimization is a simple color transform
that picks the closest color within a given tolerable deviation that minimizes
energy dissipation.
For vector graphics, Crayon applies shape transformations that slightly
enlarge or reduce the width of rectangles, lines, and polygons (or any closed
path) to reduce power dissipation in exchange of imperceptible modifications
to the geometry of the displayed shapes.
12.7 Relevant Books Available in the Engineering Library
1. Color Science: Concepts and Methods, Quantitative Data and Formulae, ISBN:
978-0471399186.
foundations of embedded systems 11
12.8 The Muddiest Point
Think about the following two questions and submit your responses through
this
link.
1. What was least clear to you in this chapter? (You can simply list the section
numbers or write a few words.)
2. What was most clear to you in this chapter? (You can simply list the
section numbers or write a few words.)
Bibliography
1. Anand, Bhojan et al. (2011). “Adaptive Display Power Management for
Mobile Games”. In: Proceedings of the 9th International Conference on Mobile
Systems, Applications, and Services. MobiSys ’11. Bethesda, Maryland, USA:
ACM, pp. 57–70 (cit. on pp. 8, 9).
2. Chang, Naehyuck, Inseok Choi, and Hojun Shim (2004). “DLS: dynamic
backlight luminance scaling of liquid crystal display”. In: Very Large Scale
Integration (VLSI) Systems, IEEE Transactions on 12.8, pp. 837–846 (cit. on
p.
8).
3. Chen, Xiang et al. (2013). “How is Energy Consumed in Smartphone Dis-
play Applications?” In: Proceedings of the 14th Workshop on Mobile Computing
Systems and Applications. HotMobile ’13. Jekyll Island, Georgia: ACM, 3:1–
3:6 (cit. on p.
8).
4. Chen, Xiang et al. (2014). “FingerShadow: An OLED Power Optimization
Based on Smartphone Touch Interactions”. In: 6th Workshop on Power-Aware
Computing and Systems (HotPower 14). Broomfield, CO: USENIX Associa-
tion (cit. on p.
9).
5. Cheng, Wei-Chung, Yu Hou, and Massoud Pedram (2004). “Power Mini-
mization in a Backlit TFT-LCD Display by Concurrent Brightness and Con-
trast Scaling”. In: Proceedings of the Conference on Design, Automation and Test
in Europe - Volume 1. DATE ’04. Washington, DC, USA: IEEE Computer So-
ciety, pp. 10252– (cit. on pp. 7, 8).
6. Cheng, Wei-Chung, Chih-Fu Hsu, and Chain-Fu Chao (2006). “Temporal
Vision-Guided Energy Minimization for Portable Displays”. In: Low Power
Electronics and Design, 2006. ISLPED’06. Proceedings of the 2006 International
Symposium on, pp. 89–94 (cit. on p.
8).
7. Choi, Inseok, Hojun Shim, and Naehyuck Chang (2002). “Low-power Color
TFT LCD Display for Hand-held Embedded Systems”. In: Proceedings of the
2002 International Symposium on Low Power Electronics and Design. ISLPED
’02. Monterey, California, USA: ACM, pp. 112–117 (cit. on p.
8).
14 phillip stanley-marbell
8. Chuang, Johnson, Daniel Weiskopf, and Torsten Möller (2009). “Energy
Aware Color S ets”. In: Computer Graphics Forum 28.2, pp. 203–211. issn:
1467-8659 (cit. on p.
8).
9. Dong, Mian, Yung-Seok Kevin Choi, and Lin Zhong (2009a). “Power Mod-
eling of Graphical User Interfaces on OLED Displays”. In: Proceedings of the
46th Annual Design Automation Conference. DAC ’09. San Francisco, Califor-
nia: ACM, pp. 652–657 (cit. on p.
8).
10. Dong, Mian, Yung-Seok Kevin Choi, and Lin Zhong (2009b). “Power-
saving Color Transformation of Mobile Graphical User Interfaces on OLED-
based Displays”. In: Proceedings of the 2009 ACM/IEEE International Sympo-
sium on Low Power Electronics and Design. ISLPED ’09. San Francisco, CA,
USA: ACM, pp. 339–342. isbn: 978-1-60558-684-7 (cit. on p.
8).
11. Dong, Mian and Lin Zhong (2011). “Chameleon: A Color-adaptive Web
Browser for Mobile OLED Displays”. In: Proceedings of the 9th Interna-
tional Conference on Mobile Systems, Applications, and Services. MobiSys ’11.
Bethesda, Maryland, USA: ACM, pp. 85–98. isbn: 978-1-4503-0643-0 (cit.
on pp. 7–9).
12. Harter, Tim et al. (2004). “Energy-aware User Interfaces: An Evaluation
of User Acceptance”. In: Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems. CHI ’04. Vienna, Austria: ACM, pp. 199–206.
isbn: 1-58113-702-8 (cit. on pp.
8, 9).
13. Iranli, Ali and Massoud Pedram (2005). “DTM: Dynamic Tone Mapping
for Backlight Scaling”. In: Proceedings of the 42nd Annual Design Automation
Conference. DAC ’05. Anaheim, California, USA: ACM, pp. 612–617 (cit. on
p.
8).
14. Land, E. H. (1959a). “Experiments in color vision.” In: Scientific American
200.5, p. 84 (cit. on pp. 3, 5).
15. Land, Edwin H (1959b). “Color vision and the natural image. Part I”.
In: Proceedings of the National Academy of Sciences 45.1, pp. 115–129 (cit. on
pp.
3, 5).
16. Land, Edwin H (1959c). “Color vision and the natural image part II”.
In: Proceedings of the National Academy of Sciences 45.4, pp. 636–644 (cit. on
pp.
3, 5).
17. Land, Edwin H (1983). “Recent advances in retinex theory and some im-
plications for cortical computations: color vision and the natural image”.
In: Proceedings of the National Academy of Sciences 80.16, pp. 5163–5169 (cit.
on pp.
3, 5).
foundations of embedded systems 15
18. Land, Edwin H (1986). “An alternative technique for the computation of
the designator in the retinex theor y of color vision”. In: Proceedings of the
national academy of sciences 83.10, pp. 3078–3080 (cit. on pp.
3, 5).
19. Land, EH (1977). “The retinex theory of color vision.” In: Scientific Amer-
ican 237.6, p. 108 (cit. on pp.
3, 5).
20. Li, Ding, Angelica Huyen Tran, and William G. J. Halfond (2014). “Mak-
ing Web Applications More Energy Efficient for OLED Smartphones”. In:
Proceedings of the 36th International Conference on Software Engineering. ICSE
2014. Hyderabad, India: ACM, pp. 527–538. isbn: 978-1-4503-2756-5 (cit.
on pp. 8, 9).
21. Li, Ding, Angelica Huyen Tran, and William G. J. Halfond (2015). “Nyx:
A Display Energy Optimizer for Mobile Web Apps”. In: Proceedings of the
2015 10th Joint Meeting on Foundations of Software Engineering. ESEC/FSE
2015. Bergamo, Italy: ACM, pp. 958–961 (cit. on p. 8).
22. Mittal, Radhika, Aman Kansal, and Ranveer Chandra (2012). “Empower-
ing Developers to Estimate App Energy Consumption”. In: Proceedings of
the 18th Annual International Conference on Mobile Computing and Networking.
Mobicom ’12. Istanbul, Turkey: ACM, pp. 317–328 (cit. on p.
8).
23. Pasricha, Sudeep et al. (2004). “Dynamic backlight adaptation for low-
power handheld devices”. In: IEEE design & test of computers 5, pp. 398–405
(cit. on p.
8).
24. Ranganathan, Parthasarathy et al. (Mar. 2006). “Energy-Aware User Inter-
faces and Energy-Adaptive Displays”. In: Computer 39.3, pp. 31–38. issn:
0018-9162 (cit. on p.
7).
25. Schuchhardt, Matthew et al. (2015). “Optimizing Mobile Display Bright-
ness by Leveraging Human Visual Perception”. In: Proceedings of the 2015
International Conference on Compilers, Architecture and Synthesis for Embedded
Systems. CASES ’15. Amsterdam, The Netherlands: IEEE Press, pp. 11–20
(cit. on p.
8).
26. Shin, Donghwa et al. (2011). “Dynamic Voltage Scaling of OLED Dis-
plays”. In: Proceedings of the 48th Design Automation Conference. DAC ’11.
San Diego, California: ACM, pp. 53–58. isbn: 978-1-4503-0636-2 (cit. on
p.
8).
16 phillip stanley-marbell
27. Stanley-Marbell, Phillip, Virginia Estellers, and Martin Rinard (2016). “Crayon:
Saving Power Through Shape and Color Approximation on Next-generation
Displays”. In: Proceedings of the Eleventh European Conference on Computer
Systems. EuroSys ’16. London, United Kingdom: ACM, 11:1–11:17. isbn:
978-1-4503-4240-7. doi:
10.1145/2901318.2901347 (cit. on p. 9).
28. Tan, Kiat Wee et al. (2013). “FOCUS: A Usable & Effective Approach to
OLED Display Power Management”. In: Proceedings of the 2013 ACM Inter-
national Joint Conference on Pervasive and Ubiquitous Computing. UbiComp
’13. Zurich, Switzerland: ACM, pp. 573–582. isbn: 978-1-4503-1770-2 (cit.
on p.
9).
29. Wang, Ji, Xiao Lin, and Chris North (2012). “GreenVis : Energy-Saving
Color Schemes for Sequential Data Visualization on OLED Displays”. Blac-
skburg, VA (cit. on p.
8).
30. Wyszecki, Gunther and WS Stiles (2000). Color Science: Concepts and Meth-
ods, Quantitative Data and Formulae. Wiley-Interscience, New York (cit. on
p.
4).
Index
GreenVis : Energy-Saving Color
Schemes for Sequential Data Vi-
sualization on OLED Displays,
8
Adaptive Display Power Management
for Mobile Games,
8, 9
An alternative technique for the com-
putation of the designator in the
retinex theory of color vision, 3, 5
Anand, Bhojan, 8, 9
Chameleon: A Color-adaptive Web
Browser for Mobile OLED Dis-
plays, 7–9
Chandra, Ranveer, 8
Chang, Naehyuck, 8
Chao, Chain-Fu, 8
Chen, Xiang, 8, 9
Cheng, Wei-Chung, 7, 8
Choi, Inseok, 8
Choi, Yung-Seok Kevin, 8
Chuang, Johnson, 8
Color Science: Concepts and Methods,
Quantitative Data and Formulae,
4
Color vision and the natural image part
II,
3, 5
Color vision and the natural image.
Part I,
3, 5
cones, 4
Crayon: Saving Power Through Shape
and Color Approximation on Next-
generation Displays,
9
DLS: dynamic backlight luminance
scaling of liquid crystal display,
8
Dong, Mian, 7–9
DTM: Dynamic Tone Mapping for
Backlight Scaling,
8
Dynamic backlight adaptation for low-
power handheld devices,
8
Dynamic Voltage Scaling of OLED Dis-
plays,
8
Empowering Developers to Estimate
App Energy Consumption,
8
Energy Aware Color Sets, 8
Energy-Aware User Interfaces and
Energy-Adaptive Displays,
7
Energy-aware User Interfaces: An
Evaluation of User Acceptance,
8,
9
Estellers, Virginia, 9
Experiments in color vision., 3, 5
FingerShadow: An OLED Power Op-
timization Based on Smartphone
Touch Interactions,
9
FOCUS: A Usable & Effective Ap-
proach to OLED Display Power
Management, 9
Halfond, William G. J., 8, 9
Harter, Tim, 8, 9
Hou, Yu, 7, 8
How is Energy Consumed in Smart-
phone Display Applications?,
8
Hsu, Chih-Fu, 8
Iranli, Ali, 8
Kansal, Aman, 8
Land, E. H., 3, 5
Land, Edwin H, 3, 5
Land, EH, 3, 5
Li, Ding, 8, 9
Lin, Xiao, 8
Low-power Color TFT LCD Display for
Hand-held Embedded Systems,
8
Möller, Torsten, 8
Making Web Applications More Energy
Efficient for OLED Smartphones,
8, 9
Mittal, Radhika, 8
North, Chris, 8
Nyx: A Display Energy Optimizer for
Mobile Web Apps,
8
Optimizing Mobile Display Brightness
by Leveraging Human Visual Per-
ception,
8
Pasricha, Sudeep, 8
Pedram, Massoud, 7, 8
Power Minimization in a Backlit
TFT-LCD Display by Concurrent
Brightness and Contrast Scaling,
7, 8
Power Modeling of Graphical User In-
terfaces on OLED Displays,
8
Power-saving Color Transformation of
Mobile Graphical User Interfaces
on OLED-based Displays,
8
18 phillip stanley-marbell
Ranganathan, Parthasarathy,
7
Recent advances in retinex theory and
some implications for cortical com-
putations: color vision and the nat-
ural image,
3, 5
Rinard, Martin, 9
Schuchhardt, Matthew, 8
Shim, Hojun, 8
Shin, Donghwa, 8
Stanley-Marbell, Phillip, 9
Stiles, WS, 4
Tan, Kiat Wee, 9
Temporal Vision-Guided Energy Mini-
mization for Portable Displays,
8
The retinex theory of color vision., 3, 5
Tran, Angelica Huyen, 8, 9
Wang, Ji, 8
Weiskopf, Daniel, 8
Wyszecki, Gunther, 4
Zhong, Lin, 7–9
