1. IAT 355 Introduction to Visual Analytics Perception May 21, 2014IAT 3551

2. Perceptual Processing  Seek to better understand visual perception and visual information processing –Multiple theories or models exist –Need to understand physiology and cognitive psychology May 21, 2014IAT 3552

3. A Simple Model  Two stage process –Parallel extraction of low-level properties of scene –Sequential goal-directed processing May 21, 2014IAT 3553 Eye Stage 1 Early, parallel detection of color, texture, shape, spatial attributes Stage 2 Serial processing of object identification (using memory) and spatial layout, action

4. Stage 1 - Low-level, Parallel  Neurons in eye & brain responsible for different kinds of information –Orientation, color, texture, movement, etc. –Arrays of neurons work in parallel –Occurs “automatically” –Rapid –Information is transitory, briefly held in iconic store –Bottom-up data-driven model of processing –Often called “pre-attentive” processing May 21, 2014IAT 3554

5. Stage 2 - Sequential, Goal- Directed  Splits into subsystems for object recognition and for interacting with environment  Increasing evidence supports independence of systems for symbolic object manipulation and for locomotion & action  First subsystem then interfaces to verbal linguistic portion of brain, second interfaces to motor systems that control muscle movements May 21, 2014IAT 3555

6. Stage 2 Attributes  Slow serial processing  Involves working and long-term memory  Top-down processing May 21, 2014IAT 3556

7. Preattentive Processing  How does human visual system analyze images? –Some things seem to be done preattentively, without the need for focused attention –Generally less than 200-250 msecs (eye movements take 200 msecs) –Seems to be done in parallel by low-level vision system May 21, 2014IAT 3557

8. How Many 3’s? 1281768756138976546984506985604982826762 9809858458224509856458945098450980943585 9091030209905959595772564675050678904567 8845789809821677654876364908560912949686 May 21, 2014IAT 3558

9. How Many 3’s? 1281768756138976546984506985604982826762 9809858458224509856458945098450980943585 9091030209905959595772564675050678904567 8845789809821677654876364908560912949686 May 21, 2014IAT 3559

10. What Kinds of Tasks?  Target detection –Is something there?  Boundary detection –Can the elements be grouped?  Counting –How many elements of a certain type are present? May 21, 2014IAT 35510

11. Examples:  Where is the red circle? Left or right?  Put your hand up as soon as you see it. May 21, 2014IAT 35511

12. Pre-attentive Hue  Can be done rapidly May 21, 2014IAT 35512

13. Examples:  Where is the red circle? Left or right?  Put your hand up as soon as you see it. May 21, 2014IAT 35513

14. Shape May 21, 2014IAT 35514

15. Examples:  Where is the red circle? Left or right?  Put your hand up as soon as you see it. May 21, 2014IAT 35515

16. Hue and Shape  Cannot be done preattentively  Must perform a sequential search  Conjuction of features (shape and hue) causes it May 21, 2014IAT 35516

17. Examples  Is there a boundary: –A connected chain of features that cross the rectangle  Put your hand up as soon as you see it. May 21, 2014IAT 35517

18. Fill and Shape  Left can be done preattentively since each group contains one unique feature  Right cannot (there is a boundary!) since the two features are mixed (fill and shape) May 21, 2014IAT 35518

19. Examples  Is there a boundary? May 21, 2014IAT 35519

20. Hue versus Shape  Left: Boundary detected preattentively based on hue regardless of shape  Right: Cannot do mixed color shapes preattentively May 21, 2014IAT 35520

21. Hue vs. Brightness  Left: Varying brightness seems to interfere  Right: Boundary based on brightness can be done preattentively May 21, 2014IAT 35521

22. Preattentive Features  Certain visual forms lend themselves to preattentive processing  Variety of forms seem to work May 21, 2014IAT 35522

23. 3-D Figures  3-D visual reality has an influence May 21, 2014IAT 35523

24. Emergent Features May 21, 2014IAT 35524

25. Potential PA Features  length  width  size  curvature  number  terminators  intersection  closure  hue  intensity  flicker  direction of motion  stereoscopic depth  3-D depth cues  lighting direction May 21, 2014IAT 35525

26. May 21, 2014IAT 35526 http://www.csc.ncsu.edu/faculty/healey/PP/

27. May 21, 2014IAT 35527 http://www.csc.ncsu.edu/faculty/healey/PP/

28. Key Perceptual Properties  Brightness  Color  Texture  Shape May 21, 2014IAT 35528

29. Luminance/Brightness  Luminance –Measured amount of light coming from some place  Brightness –Perceived amount of light coming from source May 21, 2014IAT 35529

30. Brightness  Perceived brightness is non-linear function of amount of light emitted by source Typically a power function S = aI n S - sensation I - intensity  Very different on screen versus paper May 21, 2014IAT 35530

31. Greyscale  Probably not best way to encode data because of contrast issues –Surface orientation and surroundings matter a great deal –Luminance channel of visual system is so fundamental to so much of perception We can get by without color discrimination, but not luminance May 21, 2014IAT 35531

32. Greyscale  White and Black are not fixed May 21, 2014IAT 35532

33. Greyscale  White and Black are not fixed! May 21, 2014IAT 35533

34. Luminance  Foreground must be distinct from background! May 21, 2014IAT 35534 Can you read this text?

35. Color Systems  HSV: Hue, Saturation, Value –Hue: Color type –Saturation: “Purity” of color –Value: Brightness May 21, 2014IAT 35535

36. CIE Space  The perceivable set of colors May 21, 2014IAT 35536

37. Color for Categories  Can different colors be used for categorical variables? –Yes (with care) –Colin Ware’s suggestion: 12 colors red, green, yellow, blue, black, white, pink, cyan, gray, orange, brown, purple May 21, 2014IAT 35537

38. Why 12 colors? May 21, 2014IAT 35538

39. Just-Noticeable Difference  Which is brighter? May 21, 2014IAT 35539

40. Just-Noticeable Difference  Which is brighter? May 21, 2014IAT 35540 (130, 130, 130)(140, 140, 140)

41. Weber’s Law  In the 1830’s, Weber made measurements of the just-noticeable differences (JNDs) in the perception of weight and other sensations.  He found that for a range of stimuli, the ratio of the JND ΔS to the initial stimulus S was relatively constant: ΔS / S = k May 21, 2014IAT 35541

42. Weber’s Law  Ratios more important than magnitude in stimulus detection  For example: we detect the presence of a change from 100 cm to 101 cm with the same probability as we detect the presence of a change from 1 to 1.01 cm, even though the discrepancy is 1 cm in the first case and only.01 cm in the second. May 21, 2014IAT 35542

43. Weber’s Law  Most continuous variations in magnitude are perceived as discrete steps  Examples: contour maps, font sizes May 21, 2014IAT 35543

44. Stevens’ Power Law  Compare area of circles: 1 20 May 21, 2014IAT 35544

45. Stevens’ Power Law s(x) = ax b s is the sensation x is the intensity of the attribute a is a multiplicative constant b is the power b > 1: overestimate b < 1: underestimate May 21, 2014IAT 35545 [graph from Wilkinson 99]

46. Stevens’ Power Law May 21, 2014IAT 35546 [Stevens 1961]

47. Stevens’ Power Law Experimental results for b : Length.9 to 1.1 Area.6 to.9 Volume.5 to.8 Heuristic: b ~ 1/sqrt(dimensionality) May 21, 2014IAT 35547

48. Stevens’ Power Law  Apparent magnitude scaling May 21, 2014IAT 35548 [Cartography: Thematic Map Design, p. 170, Dent, 96] S = 0.98A 0.87 [J. J. Flannery, The relative effectiveness of some graduated point symbols in the presentation of quantitative data, Canadian Geographer, 8(2), pp. 96-109, 1971] [slide from Pat Hanrahan]

49. Relative Magnitude Estimation Most accurate Least accurate Position (common) scale Position (non-aligned) scale Length Slope Angle Area Volume Color (hue/saturation/value) May 21, 2014IAT 35549

50. Color Sequence  Can you order these? May 21, 2014IAT 35550

51. Possible Color Sequences May 21, 2014IAT 35551 Grey Scale Full Spectral Scale Partial Spectral Scale Single Hue Scale Double-ended Hue Scale

52. HeatMap May 21, 2014IAT 35552

53. ColorBrewer May 21, 2014IAT 35553

54. Color Purposes  Call attention to specific data  Increase appeal, memorability  Represent discrete categories May 21, 2014IAT 35554

55. Using Color  Modesty! Less is more  Use blue in large regions, not thin lines  Use adjacent colors that vary in hue & value May 21, 2014IAT 35555

56. Using Color  For large regions, don’t use highly saturated colors (pastels a good choice)  Do not use adjacent colors that vary in amount of blue  Don’t use high saturation, spectrally extreme colors together (causes after images)  Use color for grouping and search May 21, 2014IAT 35556

57. Texture  Appears to be combination of –orientation –scale –contrast  Complex attribute to analyze May 21, 2014IAT 35557

58. Shape, Symbol  Can you develop a set of unique symbols that can be placed on a display and be rapidly perceived and differentiated?  Application for maps, military, etc.  Want to look at different preattentive aspects May 21, 2014IAT 35558

59. Glyph Construction  Suppose that we use two different visual properties to encode two different variables in a discrete data set –color, size, shape, lightness  Will the two different properties interact so that they are more/less difficult to untangle? –Integral - two properties are viewed holistically –Separable - Judge each dimension independently May 21, 2014IAT 35559

60. Integral-Separable  Not one or other, but along an axis May 21, 2014IAT 35560

61. Integral vs. Separable Dimensions Integral Separable May 21, 2014IAT 35561 [Ware 2000]

62. Shapes  Superquadric surface can be used to display 2 dimensions  Color could be added May 21, 2014IAT 35562

63. Change Blindness  Is the viewer able to perceive changes between two scenes? –If so, may be distracting –Can do things to minimize noticing changes –http://www.psych.ubc.ca/~rensink/flicker/download/ –http://nivea.psycho.univ-paris5.fr/ECS/kayakflick.gif May 21, 2014IAT 35563