Perceptually-Plausible Conversion of Images to Grayscale

Martin Čadík and Laszlo Neumann and Rafael Garcia and Antal Nemcsics
1) Department of Computer Science and Engineering, Czech Technical University in Prague, Czech Republic
2) ICREA, Barcelona, and VICOROB, University of Girona, Spain
3) Computer Vision and Robotics Group, University of Girona, Spain
4) Technical University of Budapest, Hungary

e-mails:
cadikm@fel.cvut.cz
lneumann@silver.udg.es
rafa@silver.udg.es
nemcsics.antal@t-online.hu

Keywords: color to gray transformation, color reduction, image processing, image enhancement

Abstract
Color images often have to be converted to grayscale for reproduction. Conversions which take into account local color changes are better at preserving details than global methods such as simply selecting the luminance channel. However, color changes across an image are expressed as color difference formulas with no direction, and the challenge is to find new formulas which can be used to determine an equivalent and consistent set of transitions between grays. We present two new local grayscale conversion techniques: one is based on the CIELab color difference formula, and the other on the Coloroid system and its experimental background. We obtain an inconsistent gradient field from a color image, find the closest consistent gradient field, and then integrate that to produce a grayscale image. The complexity of this method is linear in the number of pixels, making it suitable for high-resolution images; the results are pleasing and perceptually plausible.

Resources:
bibTeX entry (BIB)
Full paper (PDF)
Results
Experimental Coloroid data

An Efficient Perception-Based Adaptive Color to Gray Transformation

Laszlo Neumann and Martin Čadík and Antal Nemcsics
1) ICREA, Barcelona, and VICOROB, University of Girona, Spain
2) Department of Computer Science and Engineering, Czech Technical University in Prague, Czech Republic
3) 3Technical University of Budapest, Hungary

e-mails:
lneumann@silver.udg.es
cadikm@fel.cvut.cz
nemcsics.antal@t-online.hu

Keywords: color to gray transformation, color reduction, image processing, image enhancement

Abstract
The visualization of color images in gray scale has high practical and theoretical importance. Neither the existing local, gradient based methods, nor the fast global techniques give a satisfying result. We present a new color to grayscale transformation, based on the experimental background of the Coloroid system observations. We regard the color and luminance contrasts as a gradient field and we introduce a new simple, yet very efficient method to solve the inconsistency of the field. Having a consistent gradient field, we obtain the resultant image via fast direct integration. The complexity of the method is linear in the number of pixels, making it fast and suitable for high resolution images.

Resources:
bibTeX entry (BIB)
Full paper (PDF)
Presentation slides (PDF)
Results
Parameters