Color Harmonization

Document Sample
Color Harmonization Powered By Docstoc
					Color Harmonization
           -ACM SIGGRAPH 2006




            Speaker:李沃若
Authors

   Daniel Cohen-Or
       Tel-Aviv University
       Volume graphics
       Visualization of large environments
       Shape modeling
       Image and surface manipulation
Authors

   Olga Sorkine
       Tel-Aviv University
       Student of Daniel Cohen-Or
       Computer Graphics
   Ran Gal
Authors

   Ran Gal
       Tel-Aviv University
       Student of Daniel Cohen-Or
       Computer Graphics and Vision
       Work for Microsoft Corporation
        on the new XBOX 360
Authors

   Ying-Qing Xu
       Visual Computing Group,
        Microsoft Research Asia
       Learning based digital
        facial cartoon generation
       Computer animation
Relative Paper
   Understanding Color.   [Holtzschue 2002]

   Color design support system considering
    color harmony.   [Tokumaru et al. 2002]

   Color Index: Over 1,100 Color Combinations,
    CMYK and RGB Formulas, for Print and
    Web Media.     [Krause 2002]
Introduction
   Harmonic colors are sets of colors that are
    aesthetically pleasing in terms of human visual
    perception.
   Finds the best harmonic scheme for the image
    colors.
Color Wheel

   Describe the color
    distribution
   Use HSV color space
       H:0~360(hue)
       S:0~1 (saturation)
       V:0~1 (value)
Harmonic templates

   Sum up from experience
Some natural images
Different representation
Harmonic Schemes

1. Fit a harmonic template T to the hue
   histogram of the image.
2. Determine color shift direction.
3. Shift of color.
Fit a template

   Fine αand a template                 that
    minimize the function:




         Template      The least distant  Saturation of p
         Orientation   of hue between
                       point and template
Color shift



   C(p) : the central hue of the sector
    associated with pixel p
   w : the arc-width of the template sector
      :normalized Gaussian function
Problem

   “splitting” of a contiguous region of the
    image
Add Region Segmentation

   Fine the optimal label assignment
    minimizes the energy:

                                                Pixel’s label



         Hue distant to   color coherence
         assign edge      between neighboring
                          pixels assigned to
                          the same label.
    Min-cut / max-flow

Function can be turned
into min-cut/max-flow
problem.

Energies turn into the
cost of road.

Details see
[Boykov and Jolly 2001].
Better result
Other results
Other results
Other results
References
   BOYKOV, Y., AND JOLLY, M.-P. 2001. Interactive graph cuts for
    optimal boundary and region segmentation of objects in N-D
    images. In Proceedings of ICCV, 105–112.
   COLOR WHEEL EXPERT, 2000. http://www.abitom.com/.
   MOROVIC, J., AND LUO, M. R. 2001. The fundamentals of gamut
    mapping: A survey. Journal of Imaging Science and Technology
    45, 3, 283–290.
   PRESS, W. H., TEUKOLSKY, S. A., VETTERLING, W. T., AND
    FLANNERY, B. P. 1992. Numerical Recipes in C: The Art of
    Scientific Computing. Cambridge University Press, New York,
    NY, USA.
End

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:115
posted:12/3/2009
language:English
pages:23