[PPT] - Designing Applications that See Designing Applications that See PowerPoint Presentation

SLIDE 1

stanford hci group / cs377s

Designing Applications that See Designing Applications that See Lecture 2: Human Vision and Perception

Dan Maynes-Aminzade 10 January 2008 10 January 2008

Designing Applications that See http://cs377s.stanford.edu

SLIDE 2

R i d Reminders

Fill t th li i h t Fill out the online course sign-up sheet Assignment #1 released next Tuesday, due g y,

ne week later

R b t h k th l d f Remember to check the course calendar for the latest readings, and the course home page for announcements

10 January 2008 2 Lecture 2: Human Vision

SLIDE 3

Why Are People Taking CS377S? Why Are People Taking CS377S?

“I haven't taken any computer vision courses to date so

I haven t taken any computer vision courses to date, so

I'm interested in learning some basics.” “I've heard great things about it from previous students, d l d k and I've always wanted to take a computer vision course, but have been scared away by the theory.” “I want to build a dance interface! ” I want to build a dance interface! “It seems like a good application of my past Computer Vision and Graphics coursework, and I've always wanted t t k HCI t ” to take an HCI-type course.” “Webcams are unlike any other input device, so I'm hoping that learning to make use of them will inspire new hoping that learning to make use of them will inspire new design opportunities.” “Because Monzy's gonna rap the lectures.”

10 January 2008 3 Lecture 2: Human Vision

SLIDE 4

T d ’ G l Today’s Goals

L h h i l i k Learn how human visual processing works Compare human vision to computer vision p p Understand the limits and constraints of h i i human vision Discuss some relationships between vision, p perception, and cognition

10 January 2008 4 Lecture 2: Human Vision

SLIDE 5

O tli Outline

O i f i l t Overview of visual system Constraints of human visual processing p g Shortcuts, “hacks,” and illusions d Vision and cognition

10 January 2008 5 Lecture 2: Human Vision

SLIDE 6

A B d M d l f H Vi i A Bad Model of Human Vision

1. Eye captures scene
2. Images sent to

brain for processing

3. Brain updates

model of world

4. React and repeat loop

10 January 2008 6 Lecture 2: Human Vision

SLIDE 7

P bl ith thi M d l Problems with this Model

E t i t i i i

1. Eyes are not passive receptors; vision is an

interactive process.

10 January 2008 7 Lecture 2: Human Vision

SLIDE 8

P bl ith thi M d l Problems with this Model

E t i t i i i

1. Eyes are not passive receptors; vision is an

interactive process.

2. Processing is not serial, and reactions and

decisions are made at different stages decisions are made at different stages.

10 January 2008 8 Lecture 2: Human Vision

SLIDE 9

P bl ith thi M d l Problems with this Model

E t i t i i i

1. Eyes are not passive receptors; vision is an

interactive process.

2. Processing is not serial, and reactions and

decisions are made at different stages decisions are made at different stages.

3. We see a complex world, not just colors,

shapes, and motion.

10 January 2008 9 Lecture 2: Human Vision

SLIDE 10

Th R ti The Retina

10 January 2008 10 Lecture 2: Human Vision

(courtesy of National Eye Institute)

SLIDE 11

Th F The Fovea

10 January 2008 11 Lecture 2: Human Vision

(courtesy of Brain Connection)

SLIDE 12

B hi d th E Behind the Eyes

10 January 2008 12 Lecture 2: Human Vision

SLIDE 13

I th Vi l C t In the Visual Cortex

10 January 2008 13 Lecture 2: Human Vision

SLIDE 14

H l Hypercolumns

10 January 2008 14 Lecture 2: Human Vision

SLIDE 15

P i St Processing Streams

10 January 2008 15 Lecture 2: Human Vision

SLIDE 16

P i St Processing Streams

10 January 2008 16 Lecture 2: Human Vision

SLIDE 17

Hi h O d F ti Higher-Order Functions

10 January 2008 17 Lecture 2: Human Vision

SLIDE 18

R l ti Li it Resolution Limits

1 8 0 ° Ret ina 1 8 0 Fovea

hi h t

4 °

highest

density

f cones

Eye

10 January 2008 18 Lecture 2: Human Vision

SLIDE 19

F D Fovea Demo

10 January 2008 19 Lecture 2: Human Vision

SLIDE 20

F l E Ch t Foveal Eye Chart

10 January 2008 20 Lecture 2: Human Vision

(courtesy of Stuart Anstis)

SLIDE 21

C l t th P i h Color at the Periphery

10 January 2008 21 Lecture 2: Human Vision

(courtesy of Exploratorium)

SLIDE 22

Ph t t Di t ib ti Photoreceptor Distribution

10 January 2008 22 Lecture 2: Human Vision

SLIDE 23

Aside: Why Do Pirates Wear Eyepatches? Aside: Why Do Pirates Wear Eyepatches?

5 September 2007 23 Human Vision vs. Computer Vision

SLIDE 24

(courtesy of Jason Harrison)

SLIDE 25

(courtesy of Jason Harrison)

SLIDE 26

(courtesy of Jason Harrison)

SLIDE 27

(courtesy of Jason Harrison)

SLIDE 28

(courtesy of Jason Harrison)

SLIDE 29

(courtesy of Jason Harrison)

SLIDE 30

C t ti S l Wh l Constructing a Seamless Whole

10 January 2008 30 Lecture 2: Human Vision

(courtesy of Stuart Anstis)

SLIDE 31

S d Saccades

10 January 2008 31 Lecture 2: Human Vision

(courtesy of John M. Henderson)

SLIDE 32

E T ki Eye Tracking

10 January 2008 32 Lecture 2: Human Vision

(courtesy of Poynter Institute)

SLIDE 33

R di S d Reading Saccades

10 January 2008 33 Lecture 2: Human Vision

SLIDE 34

Th Bli d S t The Blind Spot

10 January 2008 34 Lecture 2: Human Vision

(courtesy of Peter Kaiser)

SLIDE 35

Ch hi C t Ill i Cheshire Cat Illusion

10 January 2008 35 Lecture 2: Human Vision

(courtesy of Exploratorium)

SLIDE 36

S di S i Saccadic Suppression

Y l ’ hifti You can see someone else’s eyes shifting… But when you look in a mirror, you can’t see y , y your own eyes move! Thi h l i This may help some magic tricks work – a wave with

ne hand captures your

gaze, and meanwhile you g y miss what the other hand is doing doing.

10 January 2008 36 Lecture 2: Human Vision

SLIDE 37

St d Cl k Ill i Stopped Clock Illusion

10 January 2008 37 Lecture 2: Human Vision

SLIDE 38

Sh f Sh di Shape from Shading

10 January 2008 38 Lecture 2: Human Vision

(courtesy of Dorothy Kleffner)

SLIDE 39

Sh F Sh di Shape From Shading

10 January 2008 39 Lecture 2: Human Vision

(courtesy of Dorothy Kleffner)

SLIDE 40

Sh f Sh di Shape from Shading

10 January 2008 40 Lecture 2: Human Vision

(courtesy of Dorothy Kleffner)

SLIDE 41

P O t Eff t Pop-Out Effect

10 January 2008 41 Lecture 2: Human Vision

(courtesy of Dorothy Kleffner)

SLIDE 42

R l Lif E l Real Life Example

10 January 2008 42 Lecture 2: Human Vision

SLIDE 43

R l Lif E l Real-Life Example

10 January 2008 43 Lecture 2: Human Vision

(courtesy of Susan Kare)

SLIDE 44

R l Lif E l Real-Life Example

10 January 2008 44 Lecture 2: Human Vision

SLIDE 45

R l Lif E l Real-Life Example

10 January 2008 45 Lecture 2: Human Vision

(courtesy of Stuart Anstis)

SLIDE 46

Moving Object or Changing Lighting? Moving Object or Changing Lighting?

10 January 2008 46 Lecture 2: Human Vision

(courtesy of D. Kersten)

SLIDE 47

Moving Object or Changing Lighting? Moving Object or Changing Lighting?

10 January 2008 47 Lecture 2: Human Vision

(courtesy of D. Kersten)

SLIDE 48

D th P ti Depth Perception

10 January 2008 48 Lecture 2: Human Vision

SLIDE 49

D th C Depth Cues

Bi l Binocular cues

Stereoscopic depth

P b d Perspective-based cues

Size gradient, texture gradient

Occlusion-based cues

Object overlap, cast shadows j p

Focus-based cues

Atmospheric perspective, object intensity Atmospheric perspective, object intensity

Motion-based cues

Parallax Parallax

10 January 2008 49 Lecture 2: Human Vision

SLIDE 50

P ti C E l Perspective Cue Example

10 January 2008 50 Lecture 2: Human Vision

(courtesy of Herman Bollman)

SLIDE 51

Si C E l Size Cue Example

10 January 2008 51 Lecture 2: Human Vision

SLIDE 52

At h i C E l Atmospheric Cue Example

10 January 2008 52 Lecture 2: Human Vision

(courtesy of Daniel Weiskopf)

SLIDE 53

I t it C E l Intensity Cue Example

10 January 2008 53 Lecture 2: Human Vision

SLIDE 54

B i ht L i Brightness versus Luminance

Whi h i b i ht A B? Which square is brighter, A or B?

10 January 2008 54 Lecture 2: Human Vision

(courtesy of Edward Adelson)

SLIDE 55

B i ht L i Brightness versus Luminance

Th th ! They are the same!

10 January 2008 55 Lecture 2: Human Vision

(courtesy of Edward Adelson)

SLIDE 56

Aft ff t Ill i Aftereffect Illusions

10 January 2008 56 Lecture 2: Human Vision

SLIDE 57

Aft ff t Ill i Aftereffect Illusions

10 January 2008 57 Lecture 2: Human Vision

SLIDE 58

P ti f M ti Perception of Motion

10 January 2008 58 Lecture 2: Human Vision

SLIDE 59

P ti f M ti Perception of Motion

10 January 2008 59 Lecture 2: Human Vision

SLIDE 60

M ti E t l ti Motion Extrapolation

Th “Fl h L ” Eff t The “Flash-Lag” Effect

10 January 2008 60 Lecture 2: Human Vision

SLIDE 61

M ti D t ti Motion Detection

“St i F t” Ill i “Stepping Feet” Illusion

10 January 2008 61 Lecture 2: Human Vision

SLIDE 62

D f H d Defense Hardware

Mark Leung’s “Crazy Computer Bug”

10 January 2008 62 Lecture 2: Human Vision

Mark Leungs Crazy Computer Bug

SLIDE 63

A H F B ff ? A Human Frame Buffer?

W k th t t t l We know that we construct a seamless impression of the world from many small fixations. How do these pieces get put together? How do these pieces get put together? One possible answer: the information from each fixation is accumulated in a visual buffer somewhere in the brain.

10 January 2008 63 Lecture 2: Human Vision

SLIDE 64

A H F B ff ? A Human Frame Buffer?

10 January 2008 64 Lecture 2: Human Vision

(courtesy of Jason Harrison)

SLIDE 65

W H N F B ff We Have No Frame Buffer

10 January 2008 65 Lecture 2: Human Vision

(courtesy of Jason Harrison)

SLIDE 66

Ch Bli d (1) Change Blindness (1)

10 January 2008 66 Lecture 2: Human Vision

SLIDE 67

Ch Bli d (2) Change Blindness (2)

10 January 2008 67 Lecture 2: Human Vision

SLIDE 68

Ch Bli d (3) Change Blindness (3)

10 January 2008 68 Lecture 2: Human Vision

SLIDE 69

M E l More Examples

10 January 2008 69 Lecture 2: Human Vision

SLIDE 70

M E l More Examples

10 January 2008 70 Lecture 2: Human Vision

SLIDE 71

I tt ti Bli d Inattention Blindness

10 January 2008 71 Lecture 2: Human Vision

(courtesy of Daniel J. Simons)

SLIDE 72

I tt ti Bli d Inattention Blindness

O h lf th b did t th Over half the observers did not see the person in the gorilla suit! If we are not looking for something, we

ften will not see it
ften will not see it

Instead of a complete, detailed world, we

nly see the part we are attending to

This is how magicians make things disappear This is how magicians make things disappear

10 January 2008 72 Lecture 2: Human Vision

SLIDE 73

C d T i k Card Trick

Pi k d Pick a card…

10 January 2008 73 Lecture 2: Human Vision

SLIDE 74

P t ! Presto!

I’ d d I’ve removed your card

10 January 2008 74 Lecture 2: Human Vision

SLIDE 75

St Eff t Stroop Effect

Name the COLOR of the word out loud

RED BLUE

(NOT what it spells)

RED GREEN BLUE WHITE BLUE YELLOW GREEN YELLOW YELLOW PINK YELLOW ORANGE ORANGE BLUE BLUE WHITE BLUE

10 January 2008 75 Lecture 2: Human Vision

WHITE

SLIDE 76

St Eff t Stroop Effect

Name the COLOR of the word out loud

YELLOW RED

(NOT what it spells)

YELLOW ORANGE RED YELLOW GREEN BROWN BLUE GREEN BROWN GREEN GREEN CHAIR BLUE YELLOW DONKEY LAMP YELLOW

10 January 2008 76 Lecture 2: Human Vision

LAMP

SLIDE 77

G t lt P i i l Gestalt Principles

W th ld bj t d We see the world as objects and groups, not as isolated parts Which way is the red triangle pointing?

10 January 2008 77 Lecture 2: Human Vision

SLIDE 78

P i it Proximity

W it th t l t th We group items that are close together. At right we see columns, not rows or a grid. g , g

10 January 2008 78 Lecture 2: Human Vision

SLIDE 79

Si il it Similarity

W f t t th bj t f th We prefer to group together objects of the same kind: here we see alternating rows of same-colored circles rather than columns of different-colored circles. t

10 January 2008 79 Lecture 2: Human Vision

SLIDE 80

Cl Closure

W t d t l t tt i (A) We tend to complete patterns: in (A) we see a triangle where there is none.

10 January 2008 80 Lecture 2: Human Vision

SLIDE 81

C ti ti Continuation

W lik t th ti ti f We like to see smooth continuations of shapes: here we see two lines crossing, rather than two arrowheads touching.

10 January 2008 81 Lecture 2: Human Vision

SLIDE 82

C F t Common Fate

10 January 2008 82 Lecture 2: Human Vision

SLIDE 83

C F t Common Fate

10 January 2008 83 Lecture 2: Human Vision

SLIDE 84

Bi l i l M ti i S i l Biological Motion is Special

10 January 2008 84 Lecture 2: Human Vision

SLIDE 85

F S i l Faces are Special

10 January 2008 85 Lecture 2: Human Vision

SLIDE 86

F S i l Faces are Special

10 January 2008 86 Lecture 2: Human Vision

SLIDE 87

F S i l Faces are Special

10 January 2008 87 Lecture 2: Human Vision

SLIDE 88

F F E h Faces, Faces Everywhere

5 September 2007 88 Human Vision vs. Computer Vision

SLIDE 89

5 September 2007 89 Human Vision vs. Computer Vision

SLIDE 90

A ti d F P i Austism and Face Processing

5 September 2007 90 Human Vision vs. Computer Vision

SLIDE 91

A Gi l H t Average Girls are Hot

10 January 2008 91 Lecture 2: Human Vision

(courtesy of Seed Magazine)

SLIDE 92

A F Average Faces

10 January 2008 92 Lecture 2: Human Vision

(courtesy of University of Regensburg)

SLIDE 93

“2D” P i t d R “2D” Painted Room

10 January 2008 93 Lecture 2: Human Vision

SLIDE 94

“2D” P i t d R “2D” Painted Room

10 January 2008 94 Lecture 2: Human Vision

SLIDE 95

Summary

H i l ti i l Human visual perception is complex… more than just a picture on the retina Computer vision will take a long time to catch up! catch up! But we can apply some of the same kinds of “hacks” in computer vision to simplify particular problems p p

10 January 2008 95 Lecture 2: Human Vision

SLIDE 96

R di f N t L t Reading for Next Lecture

A i i I Acquiring Images John C. Russ, Chapter 1 of The Image Processing Handbook, CRC Press, 2002.

10 January 2008 96 Lecture 2: Human Vision