Забыли?

?

# Salience-Preserving Color Removal

код для вставкиСкачать
```Color2Gray:
Salience-Preserving Color Removal
Amy Gooch
Sven Olsen
Jack Tumblin
Bruce Gooch
New Algorithm
Color
Grayscale
Problem
Volume of displayable CIE L*a*b* Colors
QuickTimeв„ў and a
Sorenson Video 3 decompressor
are needed to see this picture.
Isoluminant Colors
Color
Grayscale
Converting to GrayscaleвЂ¦
вЂў In Color Space
вЂ“ Linear
вЂ“ Nonlinear
вЂў In Image Space
вЂ“ Pixels (RGB)
вЂў Using colors in the image
вЂў Different gray for different color
вЂ“ Relative difference
вЂў Using colors in the image and their position in image space
вЂў Colors can map to same grayвЂ¦..
Luminance Channels
Problem can not be solved by
CIE CAM
Photoshop
simply97switching to a different
spaceLAB
CIE XYZ
YCrCb
вЂў Contrast enhancement &
Gamma Correction
вЂ“ DoesnвЂ™t help with isoluminant values
Photoshop Grayscale
PSGray + Auto Contrast
New Algorithm
Simple Linear Mapping
Principal Component
Analysis (PCA)
Luminance Axis
Problem with PCA
Worst case:
Isoluminant Colorwheel
Non-linear mapping
Contemporaneous Research
вЂў Rasche et al. [2005, IEEE CG&A and EG]
Color Image
Luminance Only
Rasche et al.'s Method
Goals
вЂў Dimensionality Reduction
вЂ“ From tristimulus values to single channel
Loss of information
вЂў Maintain salient features in color image
вЂ“ Human perception
Relative differences
Color Illusion by
Lotto and Purves
http://www.lottolab.org
Challenge 1:
Influence of neighboring pixels
Challenge 2:
Dimension and Size Reduction
120, 120
100
0
-120, -120
Challenge 3:
Many Color2Gray Solutions
Original
.
.
.
Algorithm Intuition
color2gray
For first pixel
Look
at DC
1
2
i = 1, j = 2
luminance
L1 L
+ 1e1,2
..
.
LL22
L1 + e1,2
For Nth pixel
Look
at DC
L1
L2
L1
L2L+2 e2,1
L2 + e2,1
Algorithm Overview
вЂў Convert to Perceptually Uniform Space
вЂ“ CIE L*a*b*
вЂў Initialize image, g, with L channel
вЂў For every pixel
вЂ“ Compute Luminance distance
вЂ“ Compute Chrominance distance
dij
вЂў Adjust g to incorporate both luminance and
chrominance differences
Color2Grey Algorithm
Optimization:
i+m
min
SS
i
j=i-m
2
( (gi - gj) - di,j )
Parameters
m : Radius of neighboring pixels
a : Max chrominance offset
q : Map chromatic difference to increases or decreases in
luminance values
m : Neighborhood Size
q= 300o
a = 10
m=2
m = 16
m= entire image
q= 49o
a = 10
m : Neighborhood Size
m = 16
m= entire image
a: Chromatic variation maps
to luminance variation
a
-a
a=5
crunch(x) = a * tanh(x/a)
a = 10
a = 25
Perceptual Distance
вЂў Luminance Distance:
DLij = Li - Lj
вЂў Chrominance Distance: ||DCij||
Problem: ||DCij|| is unsigned
Map chromatic difference to increases or
decreases in luminance values
+b*
C2
-a*
+a*
C1
Color
Space
Color
Difference
Space
+DC1,2
-
+Db*
vq = (cos q, sin q)
+
vq
q
-Da*
+Da*
-Db*
+
sign(||DCi,j||) = sign(DCi,j
.
vq )
q = 45
q = 225
Photoshop Grayscale
q = 135
q = 45
q=0
Grayscale
How to Combine
Chrominance and Luminance
dij =
DLij
(Luminance)
How to Combine
Chrominance and Luminance
DLij
(Luminance)
||DCij||
(Chrominance)
dij =
if |DLij| > ||DCij||
How to Combine
Chrominance and Luminance
DLij
if |DLij| > crunch(||DCij||)
d (a) ij =
crunch(||DCij||)
120, 120
100
a
-a
crunch(x) = a * tanh(x/a)
0
-120, -120
How to Combine
Chrominance and Luminance
d(a,q)ij =
DLij
if |DLij| > crunch(||DCij||)
crunch(||DCij||)
if DCij . nq в‰Ґ 0
crunch(-||DCij||)
otherwise
.
.
.
Grayscale
Color2Grey Algorithm
Optimization:
i+m
min
SS
i
j=i-m
2
( (gi - gj) - di,j )
If dij == DL then ideal image is g
Otherwise, selectively modulated by DCij
Results
Original
Photoshop
Grey
Color2Grey
Color2Grey
+ Color
Original
PhotoshopGrey
Color2Grey
Color2Grey+Color
Original
PhotoshopGrey
Color2Grey
Original
PhotoshopGrey
Color2Grey
Implementation
Performance
вЂў Image of size S x S
вЂ“ O(m2 S2) or O(S4) for full neighborhood case
вЂў 12.7s 100x100 image
вЂў 65.6s 150x150 image
Athlon 64 3200 CPU
вЂў 204.0s 200x200 image
вЂ“ GPU implementation
вЂў O(S2) ideal, really O(S3)
вЂ“ 2.8s 100x100
вЂ“ 9.7s 150x150
вЂ“ 25.7s 200x200
NVIDIA GeForce GT6800
Future Work
вЂў Faster
вЂ“ Multiscale
вЂў Smarter
вЂ“ Remove need to specify q
вЂў New optimization function designed to match both signed
and unsigned difference terms
вЂ“ Image complexity measures
вЂў Animations/Video
Validate "Salience Preserving"
Original
PhotoshopGrey
Color2Grey
Apply Contrast Attention model by Ma and Zhang 2003
Validate "Salience Preserving"
Original
PhotoshopGrey
Color2Grey
Thank you
вЂў SIGGRAPH Reviewers
www.color2gray.info
вЂў NSF
вЂў Helen and Robert J. Piros Fellowship
вЂў Northwestern Graphics Group
вЂў MidGraph2004 Participants
вЂ“ especially Feng Liu
вЂў (sorry I spelled your name wrong in the acknowledgements)
Original
Color2Grey
Color2Grey+Color
Original
Color2Grey
Color2Grey+Color
Original
Color2Grey
Color2Grey+Color
Original
PhotoshopGrey
Color2Grey
Original
Color2Grey
Color2Grey+Color
Original
PhotoshopGrey
Color2Grey
Original
Color2Grey
Color2Grey+Color
Photoshop Grayscale
Photoshop Grayscale
Rasche et al.
Photoshop Grayscale
Rasche et al.
Photoshop Grayscale
Rasche et al.
Photoshop Grayscale
Parameter a
a=5
a = 55
a = 15
a = 65
a = 25
a = 35
a = 45
a = 75
a = 85
a = 95
```
###### Документ
Категория
Презентации
Просмотров
9
Размер файла
8 145 Кб
Теги
1/--страниц
Пожаловаться на содержимое документа