Maintaining Color Consistency Across Non-Linear Devices

1
MAINTAINING COLOR CONSISTENCY ACROSS NON-LINEAR DEVICES Behzad Sajadi and Aditi Majumder Department of Computer Science, University of California, Irvine, Irvine, California, USA Abstract We introduce a new color transformation representation based on higher-dimensional Bezier to maintain color consistency across non-ideal devices (like commodity printers and displays). Our color transformation is more accurate and artifact-free than existing methods. It is amenable to hardware implementation making it suitable for next generation commodity devices. Objective Different input and output devices such as cameras, scanners, printers, displays, and projectors usually have different color gamuts. In order to maintain color consistency across a source and target device, a forward transformation first converts colors from the source color space to a device- independent color space (e.g. CIE XYZ). The colors in this color space are then converted to the target color space via a backward transformation. Source Device Color Space Device Independent Color Space Forward Transformat ion Backward Transformat ion Input Outpu t Target Device Color Space Color transformation steps from source device to target device • Three linearly independent primaries No channel interdependencies Channel chrominance constancy Ideal Devices Non-Ideal Devices LCD Projector LCoS Projector DLP Projector Non-Ideal Real Devices: Does not follow the above properties. 1) Sample the input RGB color space uniformly (9x9x9=729 samples) 2) Apply a 3x3 linear matrix to measured XYZ values. 3) Apply a 1D non-linear function to samples from step 2. 4) Approximate the forward transformation by three four-dimensional Bezier functions. 5) Similarly, approximate the backward transformation by three four-dimensional Bezier patches. Our Method X r Y r Z r X g Y g Z g X b Y b Z b Projected Image LAB Difference Printed Image Our Method Gamma Correction WCS Printer to DLP Projector Color Matching Color Matching Across Multiple Projectors Our Method Gamma Correction No Correction Color Matching Across Multiple Cameras sRGB to DLP Projector Color Matching Linear Interpolation (16x16x16) sRGB Image Our Method Gamma Cor. WCS Lin. 9x9x9 Lin. 16x16x16 Projected Image LAB Difference No Correct ion Gamma Correct ion Our Method (r, g, b) (X, Y, Z) (X’, Y’, Z’) B l F c T N N -1 (r, g, b) (X, Y, Z) (X’, Y’, Z’) F c T N N -1 Color gamut of projectors (shown in CIEXYZ space) are non-ideal. Reparametrizat ion Step 1 Reparametrizat ion Step 2 -1

description

Maintaining Color Consistency Across Non-Linear Devices. Behzad Sajadi and Aditi Majumder Department of Computer Science, University of California, Irvine, Irvine, California, USA. Abstract - PowerPoint PPT Presentation

Transcript of Maintaining Color Consistency Across Non-Linear Devices

Page 1: Maintaining Color Consistency  Across  Non-Linear Devices

MAINTAINING COLOR CONSISTENCY ACROSS NON-LINEAR DEVICESBehzad Sajadi and Aditi Majumder

Department of Computer Science, University of California, Irvine, Irvine, California, USA

AbstractWe introduce a new color transformation representation based on higher-dimensional Bezier to maintain color consistency across non-ideal devices (like commodity printers and displays). Our color transformation is more accurate and artifact-free than existing methods. It is amenable to hardware implementation making it suitable for next generation commodity devices.

ObjectiveDifferent input and output devices such as cameras, scanners, printers, displays, and projectors usually have different color gamuts. In order to maintain color consistency across a source and target device, a forward transformation first converts colors from the source color space to a device-independent color space (e.g. CIE XYZ). The colors in this color space are then converted to the target color space via a backward transformation.

Source Device Color

Space

Device Independent Color

Space

Forward Transformation

Backward Transformation

Input Output

Target Device Color Space

Color transformation steps from source device to target device

• Three linearly independent primaries

• No channel interdependencies

• Channel chrominance constancy

Ideal Devices

Non-Ideal Devices

LCD Projector LCoS Projector DLP Projector

Non-Ideal Real Devices: Does not follow the above properties.

1) Sample the input RGB color space uniformly (9x9x9=729 samples)

2) Apply a 3x3 linear matrix to measured XYZ values.

3) Apply a 1D non-linear function to samples from step 2.

4) Approximate the forward transformation by three four-dimensional Bezier functions.

5) Similarly, approximate the backward transformation by three four-dimensional Bezier patches.

Our Method

Xr Yr Zr

Xg Yg Zg

Xb Yb ZbProjected Image

LAB Difference

Printed Image Our Method Gamma Correction WCS

Printer to DLP Projector Color Matching

Color Matching Across Multiple Projectors

Our MethodGamma CorrectionNo Correction

Color Matching Across Multiple CamerassRGB to DLP Projector Color Matching

Linear Interpolation (16x16x16)

sRGB Image Our Method Gamma Cor. WCSLin. 9x9x9 Lin. 16x16x16

Projected Image

LAB Difference

No Correction

Gamma Correction

Our Method

(r, g, b) (X, Y, Z) (X’, Y’, Z’)

Bl

Fc

T N

N -1

(r, g, b) (X, Y, Z) (X’, Y’, Z’)

Fc

T N

N -1

Color gamut of projectors (shown in CIEXYZ space) are non-ideal.

Reparametrization Step 1

Reparametrization Step 2

-1