Colour - UBC Department of Computer Science

Colour - UBC Department of Computer Science

Colour CPSC 533C February 3, 2003 Rod McFarland Ware, Chapter 4 The science of colour vision Colour measurement systems and standards

Opponent process theory Applications The science of colour vision Receptors and trichromacy theory Red

Blue G n e re

Colour measurement systems and standards Any colour can be matched using a combination of three primaries. C rR gG bB The primaries are not necessarily red, green, and blue. Any three different colours can be used. The range of colours that can be

produced from a given set of primaries is the gamut. Colour standards CIE (Commission Internationale dclairage) Primaries chosen for mathematical properties:

do not actually correspond to colours. These virtual colours X, Y, and Z are called tristimulus values. Y is the same as luminance CIE Chromaticity Chromaticity is derived from

tristimulus values: Since x+y+z=1, just use x, y values and luminance (Y). Chromaticity diagram: x 1 y

X Y Z z X Y Z Uniform Colour Space - CIEluv

Uniform colour space: a representation where equal distances in space correspond to equal distances in perception Useful for: Specification of colour

tolerances Color coding (maximum distinction) Pseudocolour sequences to represent ordered data values CIE XYZ color space is not

uniform CIEluv is a transformation of the chromaticity diagram CIEluv does not solve all problems: Contrast effects Small colour patches: difficult to distinguish colours in the yellow-blue direction

Opponent process theory Black-white (luminance), redgreen, and blue-yellow opponents Has basis in biology and culture Should use opponent colours for coding data Properties of Colour Channels

Isoluminant / Equiluminous patterns: a colour pattern whose components do not differ in luminance Red-green and yellow-blue channels carry only about 1/3 of the detail carried by black-white. Yellow Text on a Blue Background Is fairly easy to read unless the text is

isoluminant with the background colour. As the luminance of the background becomes the same as the luminance of the text, it is very difficult to make out what the text says. So much so, that at this point I can write just about anything I want here and hardly anyone would want to put in the effort to see what it was I had written.

Other isoluminance effects Stereoscopic depth is not detectable with isoluminant colours Isoluminance in animation makes it appear to be slower than the same animation in black-and-white Shape and form are best shown using luminance:

Colour appearance Contrast Saturation Brown low

high Applications Colour selection interfaces Colour naming Natural Colour System (NCS) e.g. 0030-G80Y20 Blackness 00, intensity 30, green 80, yellow 20

Pantone, Munsell: standard colour chips Applications Colour for labelling (nominal information encoding) Distinctness A rapidly distinguished colour lies outside the convex polygon defined by the other colours in CIE

space Applications Colour for labelling (2) Unique hues: universally recognized hues (red, green, blue, yellow, black, white) should be used Contrast with background: border around objects

Applications Colour for labelling (3) Colour blindness: majority of colour-blind people cannot distinguish red-green, but most people can distinguish blue-yellow Number: only 5-10 codes easily distinguished

Applications Colour for labelling (4) Size Colour-coded objects should not be very small (about degree minimum size). Smaller objects should be more highly saturated, large colour-coded regions should have low saturation. Text highlighting should be highluminance, low-saturation.

Conventions Common usage of colours, e.g. red=stop, green=ready, blue=cold Applications Colour for labelling (5) Wares 12 recommended colours (in order of preference):

Applications Pseudocolour sequences for mapping Pseudocolouring is the practice of assigning colour to map values that do not represent colour Medical imaging Astronomical images Mapping nonvisible spectrum information to the visible spectrum (astronomy, infrared images)

Gray scale best for showing surface shape Colour best for classification Applications Colour for mapping (2) For orderable sequences, black-white, red-green, blue-yellow, or saturation (dull-vivid) sequence can be used.

For detailed data, the sequence should be based mainly on luminance. For low letail, chromatic or saturation sequences can be used. Uniform colour spaces can be used to create colour sequences where equal perceptual steps correspond to equal metric steps. Where it is important to be able to read off values from a colour map, a sequence that cycles through many colours is preferable.

Applications Colour for mapping (3) A spiral through colour space (cycling through several colours while continuously increasing in luminance) is often a good choice. Hue 0, 50,250, 45, 95

Luminance 0, 25, 50 225 Applications Colour for mapping (4) Perception: even if the sequence is smooth, people tend to see discrete colours, potentially miscategorizing data. My personal division into blue, green, yellow,

orange, red, purple: very nonlinear Applications Colour for mapping (5) Using colour for 3-D information mapping Difficult to read accurately May be used to identify regions Satellite images: regions of invisible spectrum mapped to red, green, blue channels

Applications Colour for multidimensional discrete data 5-D plot using (x, y) position, red, green, blue Possible to identify clusters Ambiguous: is a point low-red or high-green? Other methods needed to analyze clusters once identified

Rogowitz et al. How Not to Lie with Visualization Visual representation of data affects the perceived structure of the data. Enhancing data interpretation using Colour Perceptual impact of a colour is not predictable from the red/green/blue components of the

colour Mapping different aspects of colour to different data is not intuitively decodable by users. Default colour maps: rainbow Perceptual nonlinearity False contours Yellow attracts attention Guiding colour map selection

Constrain the set of colour maps available to the user based on: Data type Data spatial frequency Visualization task Other design choices made by user Representing Structure Nominal data

Object should be distinguishably different but not perceptually ordered Ordinal data Distinguishable with perceptual ordering Interval data Equal steps in data correspond to equal steps in perceived magnitude

Ratio data Zero point distinguishable in colour sequence Structure Magnitude of a variable at every spatial position Use luminance (gray scale) or saturation high

spatial frequency low Spatial Frequency luminance-based

saturation-based Segmentation Low frequency more segmentation steps can be used Highlighting Luminance-based map can be highlighted using hue variations. The highlighted

regions have the same luminance value as the rest of the map. PRAVDA Perceptual Rule-Based Architecture for Visualizing Data Accurately Part of IBMs Visualization Data Explorer (

Provides choices for colour maps based on spatial frequency, data type, and userselected goal: isomorphic (structurepreserving), segmentation, highlighting PRAVDA

Recently Viewed Presentations

  • Bad Boys and Good Girls? Patterns of Interaction and Response ...

    Bad Boys and Good Girls? Patterns of Interaction and Response ...

    Bad Boys and Good Girls? Patterns of Interaction and Response in Whole Class Teaching Myhill, Debra. (2002) Bad Boys and Good Girls? Patterns of Interaction and Response in Whole Class Teaching. British Educational Research Journal. 28. 339-352. Retrieved November 5,...
  • NETWORK EDGE INNOVATIONS Revolutionize Your xDSLs Rate and

    NETWORK EDGE INNOVATIONS Revolutionize Your xDSLs Rate and

    Times New Roman Arial Arial Narrow Wingdings MS Pゴシック Default Design Microsoft Excel Worksheet PowerPoint Presentation PowerPoint Presentation PowerPoint Presentation PowerPoint Presentation Overview How AdrenaLine Works PowerPoint Presentation PowerPoint Presentation PowerPoint Presentation PowerPoint Presentation ADSL Now - Extended ...
  • 投影片 1 -

    投影片 1 -

    18.4 Boyce/Codd Normal Form (BCNF) 18.5 Fourth Normal Form (4NF) 18.6 Fifth Normal Form (5NF) 7-Reduction E-R Model to Relational . Tables. Refer to UNIT 6, Sec. 7. Transfer your E-R model to Tables. Check each Table to see if...
  • Colonial America SS2H3: Explain the factors that shaped

    Colonial America SS2H3: Explain the factors that shaped

    How were the New England, Mid-Atlantic, and Southern Colonies founded? What is the relationship between the 3 colonial regions? Can you describe colonial life in America from the perspective of various people? How? How did the geography of the colonies...
  • Biological Image Processing

    Biological Image Processing

    Biological Image Processing Physiology and Function C428/615 Martin Jagersand With illustrations from: Tutis Vilis Talia Konkle and others as cited
  • PowerPoint Lecture Slides prepared by Barbara Heard, Atlantic

    PowerPoint Lecture Slides prepared by Barbara Heard, Atlantic

    Retropulsion: The pyloric end of the stomach acts as a pump that delivers small amounts of chyme into the duodenum, simultaneously forcing most of its contained material backward into the stomach. 2 Propulsion: Peristaltic waves move from the fundus toward...
  • Random Projection for High Dimensional Data Clustering: A ...

    Random Projection for High Dimensional Data Clustering: A ...

    A single run of clustering consists of applying random projection to the high dimensional data and clustering the reduced data using EM. Multiple runs of clustering are performed and the results are aggregated to form an n n similarity matrix....
  • Trey Research

    Trey Research

    Trey Research. Violence in the Media. Results: Too Much Media Violence. Nearly 90% say TV shows are too violent. All say primetime TV shows are worst. More than 75% say violence is visible in commercials. About 50% say TV news...