Re: Color difference equations
Re: Color difference equations
- Subject: Re: Color difference equations
- From: Mark McCormick-Goodhart <email@hidden>
- Date: Fri, 17 Jul 2009 16:06:40 -0400
Anyone interested in this thread on the precision and validity of
color difference equations, when applied in the context of real
photographic images, may also want to take a look at two papers about
the I* metric. The papers are entitled "An Introduction to the I*
Metric" and "A “Retained Image Appearance” Metric For Full Tonal
Scale, Colorimetric Evaluation Of Photographic Image Stability". They
are the 2nd and 3rd PDF documents found on the following page at my
website:
http://www.aardenburg-imaging.com/documents.html
Any of the various flavors of dE are designed to address perceptual
differences in two isolated side-by-side colors. Great for matching
paint and textile colors. Also good for image process control as noted
by others. However, real images are complex arrangements of colors
such that our judgement of color and tonal accuracy also requires an
additional dimension of near neighbor contrast not considered by any
of the color difference models. Moreover, humans weight low chroma
colors as much more important to scene color balance than high chroma
colors. In other words, it is not just about perceiving a visual
difference, it''s about interpreting what the observed difference
means in the context of the image information content. Although some
of the color difference models have some chroma weighting, it is not
at all like what is needed to deal with the issue in a complex image.
The I* color function treats chroma weighting very differently than dE
models. Moreover, the I*tone function handles both lightness and
contrast. I* metric scaling is a percentile scale with 100% being a
perfect match between reference and comparison. 0% I* color values
mean no color accuracy left, and 0% I* tone means no contrast (hence
no spatial content left) while negative values for I* color mean that
false color now exists (e.g., a skintone turned green or a blue sky
turned magenta). and negative I* tone values mean a contrast inversion
(e.g., neighboring tones reversed like a photographic negative).
best regards,
Mark McCormick-Goodhart
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden