A New Efficient Real-Time Global Tone Mapping Method | TU Wien

Information

Publication Type: Master Thesis
Workgroup(s)/Project(s):
- Rendering and Modeling
Date: June 2006
First Supervisor:
- Werner Purgathofer
- Laszlo Neumann

Abstract

In computer graphics, visualization, digital imaging and specially in digital pho- tography, tone mapping plays an important role. The contrast of an image is ¯nite and display device dependent. It is typically around 4-6 stops according to photo- graphical tradition in log2 scale. The contrast of the original scene however is often signi¯cantly greater than that so that the bright or dark parts of the scene will have no details in the image. A classical photography pipeline doesn't manipulate the image after an automatic exposure. In this work we give an overview of color science, human vision and display de- vices. We present known global tone mapping methods and some e±cient adaptive HDR methods. The main focus of the work is to ¯nd the optimum contrast window for global tone mapping and to decide if there are better global and local mapping methods than the widespread mean value mapping. This work deals with the generalization of the minimal information loss method and the empirical search for optimal parameter combinations for the method. It is combined, in an innovative way, with the max(r,g,b) function (instead of lu- minance) introduced in incident light metering for computer graphics, or quasi- irradiance principle. These two methods yield visually pleasant results high con- trast images, e.g. for the always critical back light or images containing exterior and interior parts simultaneously. Since the method works in real time, after further improvements it could be easily included in Digital Cameras.

Additional Files and Images

paper

Weblinks

No further information available.

BibTeX

@mastersthesis{spielmann-2006-ner,
  title =      "A New Efficient Real-Time Global Tone Mapping Method",
  author =     "Georg Spielmann",
  year =       "2006",
  abstract =   "In computer graphics, visualization, digital imaging and
               specially in digital pho- tography, tone mapping plays an
               important role. The contrast of an image is ¯nite and
               display device dependent. It is typically around 4-6 stops
               according to photo- graphical tradition in log2 scale. The
               contrast of the original scene however is often
               signi¯cantly greater than that so that the bright or dark
               parts of the scene will have no details in the image. A
               classical photography pipeline doesn't manipulate the image
               after an automatic exposure. In this work we give an
               overview of color science, human vision and display de-
               vices. We present known global tone mapping methods and some
               e±cient adaptive HDR methods. The main focus of the work is
               to ¯nd the optimum contrast window for global tone mapping
               and to decide if there are better global and local mapping
               methods than the widespread mean value mapping. This work
               deals with the generalization of the minimal information
               loss method and the empirical search for optimal parameter
               combinations for the method. It is combined, in an
               innovative way, with the max(r,g,b) function (instead of lu-
               minance) introduced in incident light metering for computer
               graphics, or quasi- irradiance principle. These two methods
               yield visually pleasant results high con- trast images, e.g.
               for the always critical back light or images containing
               exterior and interior parts simultaneously. Since the method
               works in real time, after further improvements it could be
               easily included in Digital Cameras.",
  month =      jun,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Institute of Computer Graphics and Algorithms, Vienna
               University of Technology ",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2006/spielmann-2006-ner/",
}