Optically adjustable display color gamut
in time-sequential displays using
LED/Laser light sources
Displays
vol. 27, 2006
Moon-Cheol Kim
School of Electrical Engineering and Computer Science
Kyungpook National Univ.
Abstract
‹
Development of various wide color gamut displays
– Problem
• Color gamut differences between wide gamut displays and the
standard definitions
• Color reproduction is not equal to original images
– Overcome method in conventional systems
• Gamut mapping algorithm
– Manipulation at only the video signal that would reduce the signal
dynamic range and producing quantization error of video signal
‹
Proposed method
– A new optical Color Gamut Processing
– Realization within a display gamut without any signal artifacts
with regard to perfect color match for the various standards
2 /23
Introduction
‹
Wide color gamut displays
– Having different color gamut compared to the standard color
gamut of CRT
– Multi-primary display (more than three primary colors)
– Display with highly saturated pure
primary colors (LED, Laser)
‹
Mismatch of color gamuts
– Causing color distortion on reproduced images
– Appling color transformation
• Reduction of the dynamic range of video
signal after the color transformation
Fig.1. Color gamuts of LED and
Laser displayin comparison with
standard sRGB(CRT) on CIE-xy
chromaticity diagram
3 /23
‹
MCGP (Multi-standard Color Gamut Processing)
– A simple optical process which composes new primary colors by
an optical additive mixture of original primary colors in a timesequential display system
– A luminance increment
• The display of high luminance is preferred for the better image
contrast
– The exact matching of the color gamut of commercial displays in
serial production line
‹
DCGP (Dynamic Color Gamut Processing)
– A new scene adaptive color gamut
• Creation by analyzing the color distribution for each video frame
– Increasing the image contrast
• Inversed proportion to the size of scene color gamut
4 /23
Multi-standard Color Gamut
Processing
‹
The basic idea
– A color inside the gamut triangle
• Reproduction by mixing of the certain light portion of three primary
color on CIE − xy chromaticity diagram
‹
The new color gamut
– Composition by optical mixture of
original display color gamut
Pi ' = ai ∗ P1 + bi ∗ P2 + ci ∗ P3
for i = 1, 2, 3,
(1)
Where Pi = ( X i , Yi , Z i ) : the tristimulus values of
Fig. 2. Composing of new primary
the corresponding primary color
colors by optical additive mixture
ai , bi , and ci : the modulation factors
of three primary colors
for light mixing
5 /23
– Full matrix form
M' = G ⋅ M
 X 1' Y1'
 '
 X 2 Y2'
 X' Y'
 3 3
Where
Z1'   a1
 
'
Z 2  =  a2
Z 3'   a3
b1
b2
b3
c1   X 1 Y1

c2  ⋅  X 2 Y2
c2   X 3 Y3
(2)
Z1 

Z2 
Z 3 
G : the required factors of the light modulation
of the original primary colors
– To get a modulation gain matrix G
G = M ' ⋅ M -1
(3)
– The display forward matrix M
• Dividing into constant primary coordinate matrix
adjustable normalizing matrix N
P and an
6 /23
M=P⋅N
Where
 x1

P =  x2
x
 3
z1 
 n1


z 2  and N =  0
0
z3 

y1
y2
y3
0
n2
0
0

0
n3 
(4)
n1 −n3 : the modulation coefficients of three primary colors
in order to set an individual display white
– To maximize the white luminance after the MCGP
G M = G / Max (a1 , b2 , c3 )
 a1n b1n c1n 


=  a2 n b2 n c2 n 
a

b
c
3n
2n 
 3n
(5)
– The rising ratio of the display luminance before and after MCGP
∆Y =
∑Y / ∑Y
'
i
i =1, 2 , 3
i =1, 2 , 3
i


=  Y1 ⋅ ∑ ai + Y2 ⋅ ∑ bi + Y3 ⋅ ∑ ci  / (Y1 + Y2 + Y3 )
i =1, 2 , 3
i =1, 2 , 3 
 i =1, 2,3
(6)
7 /23
Dynamic Color Gamut Processing
‹
The basic idea
– An optimized control of display color gamut in connection with a
color gamut of video frames
– Variety of color gamut of each video frame
– No covering the full range of the display color gamut
Fig. 3. Schematic diagram for the
determination of scene color gamut
8 /23
‹
Proposed DCGP
– Controlling the display color gamut depending on a measured
scene color gamut
– Increasing the display luminance
– The simplified block diagram of DCGP
• Computing a new primary system for every video frame
• Appling to maintain the original colors of image pixels on the new
primary color system as color transformation
Fig. 4. Functional block diagram of DCGP
9 /23
‹
Precessing
– Linearization for nonlinear RGB NL by the inverse tone curve
characteristic
– Transformation of the linear sRGB by matrix operation M s to the
tristimulus signal XYZ (first part)
F = M s ⋅ Cs
with F = ( X , Y , Z )T ,
Cs = ( Rs , Gs , Bs )T
(7)
– Transformation to the CIE − xy chromaticity values
 x  X 
  =   / ( X + Y + Z )
 y  Y 
– Getting a scene color gamut (triangle Pr' , Pg' and Pb' )
(8)
• Determining three independent primary-lines ( L , L , L )
RG
GB
BR
Pr = ( xr , yr ), Pg = ( x g , y g )
x − xr
y − yr
=
x g − xr y g − y r
(9)
10 /23
• Determining a minimum distance d between the primary-line and a
test pixel chromaticity ( x p , y p )
– An example between the line L RG and the test pixel ( x p , y p )
d=
( x p − xr )( xg − xr ) − ( y p − yr )( y g − yr )
( x g − xr ) + ( y g − y r )
2
2
(10)
• Determining new primary-lines (L' RG . L'GB , L' BR )
– Passing the three nearest points and paralleling to the (L RG . L GB , L BR )
y − y p , RG
x − x p , RG
y − y p ,GB
x − x p ,GB
=
=
y g − yr
x g − xr
yb − y g
;
y − y p , RB
x − x p , RB
y B − yr
=
;
x B − xr
(11)
xb − x g
11 /23
• The final new primary points ( Pr' ( xr' , yr' ), Pg' ( xg' , y g' ), Pb' ( xb' , yb' ))
– Solving as cross points of the corresponding two primary lines
 xr'
 '
PD =  x g
 x'
 b
yr'
y g'
yb'
z r' 

'
zg 
zb' 
(12)
– The display forward model for the new primary system
• With the matrix PD and a defined target white FW = (X W , YW , Z W )
F = PD ⋅ N ⋅ C D = M D ⋅ C D
with F = ( X , Y , Z ) T , C D = ( RD , GD , BD ) T
and N = (( N r , 0 , 0), (0 , N g , 0), (0 , 0, N b ))
(13)
where normalizing matrix Ν : defined white Fw at the maximum
display control vector C D = (1,1,1)
12 /23
• After finding the new display matrix M D , same process with the
process in MCGP (second part)
• The final modulation gain matrix for DCGP
– Use to light modulation circuit in display hardware in order to
modulate the primary colors as explained in MCGP
 a1D

G D =  a2 D
a
 3D
b1D
b2 D
b3D
c1D 

c2 D 
c3D 
(14)
• Color transformation for the same color look of displayed images
before and after DCGP
– Doing by 3x3 matrix transform M T (Eqs. (7) and (13))
CD = M D−1 ⋅ M S ⋅ CS = M T ⋅ CS
with CD = ( RD , GD , BD ) T , CS = ( RS , GS , BS ) T
⇒ M T = M D−1 ⋅ M S
(15)
13 /23
– DCGP at all together (first part and second part)
• Changing color gamut according to input images
• Raising display luminance by using surplus intensity
– If MCGP and DCGP are applied at the same time
• MCGP
– Static setting display color gamut according to the color gamut
standards
•
DCGP
– Dynamical controlling the gamut of display for an individual scene
color gamut
• Unifying two modulation matrixes G M and G D to the total
modulation gain matrix G T
GT = GM ⋅GD
(16)
14 /23
Practical simulation
‹
A principal diagram
– Separation of the RGB primary colors at time
– Presentation of individual color as sequential mixing by
modulating the primary colors
Fig. 5. (a) Principal driving
method of a sequential color
display and the conventional
driving method (b) PWM and
(c) AM driving for optical mixing
of primary colors
15 /23
– Methods for the light modulation
• PWM (pulse width modulation) (Fig. 5b)
• AM (amplitude modulation) (Fig. 5c)
– The coefficient indexes al – c3 (Fig. 5c)
• The coefficients of the modulation matrixes
• A mixture ratio of the physical light intensity
• Conversion from light intensity to the electric current
– By a predetermined relationship
Fig. 6. Typical nonlinear light intensity
characteristic of power LEDs in function
of driving currents
16 /23
‹
The MCGP simulation
– Using LED − projection display based on a time-sequential
DLP − technology
– Using gamut standard of Adobe − RGB and sRGB as target gamut
Table 1. Chromaticity values of primary colors for an LED display, two standard definitions,
and the luminance increasing factor by applying MCGP
Fig. 7. Color gamut of the LED − projection
display in comparison with the two gamut
standards of Adobe − RGB and sRGB
17 /23
– First step
• Obtaining the two matrix data (M, M ' ) (Eq. (2))
– From the primary coordinates of the LED display the target color space
sRGB or Adobe − RGB, and a predefined white point
– D65 white for both gamut cases
– Second step
• Computation of the matrix G M = G n for sRGB target gamut
• Modulating the primary colors of the LED display
 0.412

M sRGB =  0.212
 0.019

 0.652

G M,sRGB =  0.309
 0.089

0.357 0.180 

0.715 0.072 
0.119 0.950 
0.065 0.012 

0.957 0.039 
0.028
1 
 0.644 0.178 0.127 


M LED =  0.272 0.695 0.031 
 0.004 0.087 0.996 


(17)
– Perfect matching from the LED display to sRGB standard system
Increasing display luminance of about 29.5% (Table 1)
18 /23
‹
The DCGP simulation
– Obtaining the test images typical scenes from DVD titles
– Calculating corresponding color gamuts of the scenes
– Creating new display gamut for the scenes using the matrix G D
– Displaying the images with an increased scene luminance and
same chromaticity values of image pixels on the display with the
scene color gamut
Fig. 8. Original simulation images
19 /23
Fig. 9. Determined scene color gamuts for the corresponding images 1-8
20 /23
– Result of DCGP
• For the visualization of the result of DCGP simulation
– Using the images with half-luminance
• Having increase in the range of 25-71.4% depending on scene
gamuts for the scene luminance
• Having average increasing ratio for typical DVD titles in the range of
about 30%
Fig. 11. Results of DCGP simulation
21 /23
Fig. 10. Input images with 50% luminance reduction from Fig. 8
and output images of DCGP simulation with these input images
22 /23
Conclusions
‹
Proposed methods
– MCGP
• Application to the wide color gamut display for the perfect match
of the display gamut to individual standard color gamut without
any video processing
• Increasing the display luminance without any reduction of the
signal dynamic
– DCGP
• Adaptation of the display gamut to scene color gamuts of
corresponding input video frames in order to enhance the scene
luminance without color distortion
• Providing more scene luminance for a low luminance display
– The scene luminance for sRGB color reproduction
• Increase by about 1.6 times more than the display
23 /23