Information Visualization

1
Information Visualization

• Kathryn Summers
• 2004

2
Data visualization

• How to design attention-grabbing features
• How to code data so that patterns become
  perceptible

3
Data visualizationwhy do it?

• Facilitates comprehension of large amounts of
  data
• Allows perception of unanticipated properties
  (patterns)
• May expose problems with the data, contributing
  to quality control
• Facilitates understanding large and small scale
  features of the data
• Support hypothesis formation

4
Stages of data visualization

• Collecting and storing the data
• Preprocessing data to transform it into something
  understandable
• Display hardware and software (graphics
  algorithms) to produce visualization
• Reception by human perceptual and cognitive system

5
Perceptual processing

• Stage 1pre-attentive processing (20 of cortical
  neurons)
• Parallel processing (fast)
• Data-driven
• Color, motion, elements of form, contour,
  orientation, texture
• Short-term memory
• Stage 2
• Visual field segmented into regions and
  two-dimensional patterns

6
Stage 3 perceptual processing

• Sequential processing (slow)
• Action-oriented processing
• Semiotic processing
• Goal-driven
• Involves working and long-term memory

7
Eye movements

• Saccadic movements
• movement takes 20-100msec, dwell period
  200-600msec
• Ballisticcant be interrupted
• Saccadic suppressionmay not see events during
  saccade
• Smooth pursuit movements
• Convergent movements
• eyes converge if object moving closer, diverge if
  moving away
• Accommodation200 msec to refocus after movement

8
Field of view

• Varies
• Larger if low density environment
• Smaller if high density environment
• Narrows as cognitive load increases
• Tunnel vision in emergencies

9
Pre-attentively processed characteristics

• Orientation
• Size
• Shape
• Convexity
• Concavity
• Added box around object
• (See figure 5.5)

• Pre-attentive processing takes 10 msec or less
  per item (time increases with variety)
• Non-pre-attentive processing takes 40 msec or
  more per item

10
Specialized vision cells

• Specialized cells for detecting
• Orientation and size, with luminance
• Color (two types of signal)
• Stereoscopic depth
• Motion
• Gabor receptive fields
• Orientation
• Scale
• Contrast
• Hue (added to Gabors model)

11
Human visionsharpening

• Brain is most sensitive to
• patterns of differences
• Changes over time
• Brain uses the ways in which groups of cells fire
  to increase resolution of differences
• Retinal ganglion cells respond differently to
  on-center and off-center stimulation
• On-center stimulation?more neuron pulses
• Off-center stimulation?depressed pulses
• The result is heightened perceived contrast
• pg 77 and 83

12
Conjunctionssome are pre-attentive, most are not

• Space and color
• Space and shape
• Depth and color
• Depth and movement

• Convexity and color
• Motion and shape

13
Integral and separable dimensions (figure 5.25)

• Integral dimensions cause interference, but can
  speed processing with redundant coding
• Separable dimensions dont cause interference,
  but dont speed processing with redundant coding

14
Vision and computer monitors

• Center of monitor at typical distance stimulates
  gt50 of the visual processing mechanisms in the
  brain
• Resolution about 40 pixels/cm human visual
  acuity (for lines) accurate to about 1/10 pixel
• Monitor cannot provide depth of focus info may
  confuse human spatial processing systems
• Red and blue phosphors resolve at different
  distances, so monitor blue is usually out of
  focus (solution add some red green)

15
Human interrupt signals

• Warnings, routine change of status, patterns of
  events
• Channels for informationdisplay windows, dials,
  speakers
• Eventssignals on channels
• Expected costcost of missing an event

16
Characteristics of human interrupts

• Should be easily perceived even outside locus of
  attention
• Should continue to remind user even if ignored
• Should not be annoying
• Should reflect appropriate (varying) levels of
  urgency

17
Channel monitoring behavior

• Growth of uncertainty about state of channel
• Cost of sampling
• Proximity of channels
• Frequency of events (channels with infrequent
  events may get oversampled or forgotten)
• Learned scan patterns (habit)

18
Implications for interrupt design

• Events can be missed during a saccade
• Peripheral vision is color blind
• Too rapid motion is so hard to ignore its likely
  to be annoying
• Motion is already overused (banner pop-up
  blindness)
• Use smooth motion, or repeated appearance/disappea
  rance of object, or even auditory cues

19
Opponent process color theory

• Luminance is based on input from all three cones
• Red-green is based on the difference between
  long- and middle-wavelength cones (color
  blindness can come from lack of long or medium
  cones)
• Yellow-blue is based on the difference between
  short wave-length cones and the sum of the other
  two

Long (red)
luminance
R-G
Medium (green)
Y-B
Short (blue)
20
Properties of color channels

• Luminance, or black-white channel, carries the
  most detail
• The R-G and Y-B channels only carry 1/3 as much
  detail
• Stereo perception, and perceived speed of motion,
  and even shape are based mostly on luminance
• We adapt to colored lightso that the neutral
  point for white shifts (for example, indoor
  light is yellow, but we still perceive white as
  white)

21
Color in interface design

• Some colors seem to be cross-cultural black,
  white, red, yellow, green, blue are the most
  standard brown, pink, purple, orange, and gray
  are the next most common
• Most people can reliably identify pure yellow
• There seem to be two pure greens, one at 514 nm
  and one at 525 nm (about 1/3 of people)
• Appearance of colors will change in context

22
Color in interface design

• Use high-saturation colors to code small objects
• Use low-saturation colors for large areas
• Add a white or black border to reduce contrast
  effects
• Yellow-blue contrast less vulnerable to color
  blindness
• Limit the number of color codes to 5-10

23
Next Week

• Motion, patterns, and data objects

24
Field of view and virtual reality
25
Spatial contrasts

• Spatial frequency
• Orientation
• Contrast
• Phase angle (lateral displacement of pattern)
• Area covered by the pattern