High Color-Fidelity Imaging System Using B/W Digital ... · High Color-Fidelity Imaging System...

33
1 High Color-Fidelity Imaging System Using B/W Digital Camera and Rotary Color Filters Fuji Photo Film Hitoshi Urabe Mitsubishi Electric Hiroaki Sugiura Note This research was organized in part by the New Energy and Industrial Technology Development Organization, Japan(NEDO).

Transcript of High Color-Fidelity Imaging System Using B/W Digital ... · High Color-Fidelity Imaging System...

1

High Color-Fidelity Imaging System Using B/W Digital Camera and

Rotary Color Filters

Fuji Photo Film Hitoshi UrabeMitsubishi Electric Hiroaki Sugiura

NoteThis research was organized in part by the New Energy and Industrial Technology Development Organization, Japan(NEDO).

2

Part 1: To develop NEDO-DSC to aquire a high-resolution

colorimetric image to reproduce an actual scene.

Part 2: Its applications

3

Scene

Capturing

Part 1 : Backgrounduse case : RAW-DSC for professional user

Appearance conv.

Mapping to device

Monitor Print

XYZ X’Y’Z’

RGBorCMYColorimetric color reproduction

4Color Filter B/W DSC

Capturing deviceHigh accurate and high fidelity color capturing system

B/W DSC with rotary color filters

5

Transmittance spectra of the color filters

0

0.2

0.4

0.6

0.8

1

380 480 580 680 780wavelength(nm)

trans

mitt

amce

Filter1

Filter2

Filter3

6

0

0.25

0.5

0.75

1

380 480 580 680 780w avelength (nm)

sens

itivity

)(λx

)(λy

)(λz)(λx′

)(λy

)(λz′

7

Evaluation of the spectral sensitivities

Neugebauer’s quality factor

Measure of Goodness by Vora and Trussell

23

1

2 saqi

i∑=

=

33

1∑=

≡i

iqν

8

Q-factor andν

DSC R G B

NEDO-DSC 0.999 0.992 0.983 0.991

DSC(Company A) 0.776 0.996 0.874 0.882

DSC(Company B) 0.813 0.941 0.848 0.867

ν

9

Color registration

Color Filter

Optical Axis

Actual ray of light

10

Registration check

11

Registration error

Type Filter X(pixel) Y(pixel) Rotation(deg.)

ARB

0.025514-0.024398

0.0994840.014183

0.00100.0005

ARB

-0.033403-0.074217

0.038653-0.020859

-0.0001-0.0002

BRB

0.012775-0.067492

0.029657-0.016036

-0.0005-0.0005

BRB

-0.032086-0.087687

0.0118660.055662

-0.0002-0.0001

12

Data processing procedure

RAW images

Honeycomb interpolation

Interpolated image

XYZ conv.

Calibration data

XYZ image

13

Calibration method

Scene :Macbeth&NCS chart& white plate

Spectroradiometer

B/W DSC with R-CF

Chart:Plate:

nnn BGR ,,www BGR ,,

Tristimulus values

Chart:Plate:

nnn ZYX ,,www BGR ,,

Image data

1

1

1

1

1

1

1

''''''G''' −

⎟⎟⎟

⎜⎜⎜

⎟⎟⎟

⎜⎜⎜

⎛=

wn

wn

wn

wn

wn

wn

BBBGGRRR

ZZZYYYXXX

A・・・

・・・

・・・

・・・

・・・

・・・

⎪⎩

⎪⎨

−−=−−=−−=

)/()(')/()(')/()('

γγββαα

wnn

wnn

wnn

BBBGGGRRR

3×3 MTX:Correction terms:

Aγβα ,,

Calibration data

14

Spectra of the illuminants

0

2

4

6

8

10

12

380 480 580 680 780w avelength (nm)

radi

ant e

nerg

y (Y

=100

)1: FL40SSW/37(TOSHIBA)2: XC-500AF(SERIC)3: PRF500WD(IWASAKI)

15

Colorimetric accuracy of NEDO-DSC

Illuminant

1(FL40SSW/37) 1.77 0.38 6.27

2(XC-500AF) 1.86 0.54 6.25

3(PRF500WD) 2.08 0.72 7.44

EΔ gEΔ maxEΔ

16

Comparison between NEDO-DSC and Spectro-radiometer

-60

-40

-20

0

20

40

60

80

-60 -40 -20 0 20 40 60

a*

b* PR650

NEDO-DSC

17

Comparison between NEDO-DSC and commercially available DSC(S2Pro)

-60

-40

-20

0

20

40

60

80

-60 -40 -20 0 20 40 60

a*

b* NEDO-DSC

S2Pro

18NEDO-DSCDSC(S2Pro)

sRGB image appearance

19

Conclusion

NEDO-DSC has good results as follows.(1)Average delta-E = 1.86

to MacBeth 24 and NCS 40 colours andAverage delta-E = 0.54 to MacBeth 6 grays under XC-500AF(SERIC) illuminant.

(2)Much better than commercially available DSC(S2Pro) in terms of Colorimetric reproduction.

20

Part 2 : Its applications.Actual scene-referred colour space

using NEDO-DSC

21

Color gamut of actual scenesProceduresScene : 5 scenes were captured using NEDO-DSC.

Each scene has extremely saturated objects including plastics, textiles and so on.

Calibration targets : Macbeth(24), NCS(40) and W-plate

Calculation : Most saturated colors of each scene were calculated in L* in every deltaL*=10 and

Hue in every 10 degree.

Then, outer most boundary(Maximum C* of 5 scenes) were calculated and plotted in next slides.

22

Five scenes

23

Actual scene-referred color gamut(D65)

:OptimalColor:Target5

:AdobeRGB

Lab color space (L*= 20)

-160

-120

-80

-40

0

40

80

120

160

-200 -160 -120 -80 -40 0 40 80 120 160

a*

b*

O ptim alC olor

Target5

AdobeRG B

Lab color space (L*= 50)

-160

-120

-80

-40

0

40

80

120

160

-200 -160 -120 -80 -40 0 40 80 120 160

a*

b*

O ptim alC olor

Target5

AdobeRG B

Lab color space (L*= 80)

-160

-120

-80

-40

0

40

80

120

160

-200 -160 -120 -80 -40 0 40 80 120 160

a*

b*

O ptim alC olor

Target5

AdobeRG B

24

Actual scene-referred color gamut(D65)

:OptimalColor:Target5

:AdobeRGB

C -R (190°~40°)

0

10

20

30

40

50

60

70

80

90

100

-200 -150 -100 -50 0 50 100 150 200

C *

L*

O ptim alC olor

Target5

AdobeRG B

M -G (330°~150°)

0

10

20

30

40

50

60

70

80

90

100

-200 -150 -100 -50 0 50 100 150 200

C *

L*

O ptim alC olor

Target5

AdobeRG B

B-Y (310°~100°)

0

10

20

30

40

50

60

70

80

90

100

-200 -150 -100 -50 0 50 100 150 200

C *

L*

O ptim alC olor

Target5

AdobeRG B

25

Part 2 : Its applications.In case of transmissive and

fluorescent objects

26

Eight semi-ransmissive scenes

27

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

380 420 460 500 540 580 620 660 700 740 780

nm

inten

sity

ST_No13ST_No15ST_No16WhitePlate

Spectral distribution of semi-transmissive objects

28

Part 2 : Its applications.Actual scene-referred colour space

using NEDO-DSC

29

Spectral distribution of UV illuminant

0

50000

100000

150000

200000

300 400 500 600 700 800

nm

Relat

ive In

tens

ity

With UVWithout UV

30

Four fluorescent scenes

31

0.0

0.5

1.0

1.5

2.0

2.5

380 420 460 500 540 580 620 660 700 740 780

nm

inten

sity

No14No15WhitePlate

Spectral distribution of fluorescent objects

32

Future Challenges

To define actual scene-referred colour gamut

To develop estimating method of colorimetric characteristics of high-end RAW-DSCs

AcknowledgementThis research was organized in part by the New Energy and Industrial Technology Development Organization, Japan(NEDO).

33

Thank you for your attention.