Top Banner
1 THE VISUAL SYSTEM Visual Perception Aditi Majumder, UCI Slide 2 Aditi Majumder, UCI Visual System Eye - sensor Lateral geniculate nucleus Striate cortex Striped Appearance Extra striate cortex
32

THE VISUAL SYSTEM - Donald Bren School of Information and Computer

Feb 03, 2022

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

1

THE VISUAL SYSTEM

Visual PerceptionAditi Majumder, UCI

Slide 2 Aditi Majumder, UCI

Visual System

Eye - sensorLateral geniculate nucleusStriate cortex

Striped AppearanceExtra striate cortex

Page 2: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

2

Slide 3 Aditi Majumder, UCI

Visual System

Slide 4 Aditi Majumder, UCI

Visual System

Page 3: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

3

Eye : The Sensor

Slide 6 Aditi Majumder, UCI

Eye : Structure

Page 4: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

4

Slide 7 Aditi Majumder, UCI

Eye : Accommodation

Accommodation : Flexible focusing abilityCornea – 80% of focusing abilityLens – 20% adaptable focusing ability

Focus at one depth at a timeLimitation

Depth of fieldNear and far point

Slide 8 Aditi Majumder, UCI

Myopia/Presbyopia

Far and near point is different for different individuals

If near point is far: presbyopiaIf far point is near: myopia

Page 5: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

5

Slide 9 Aditi Majumder, UCI

Iris

Muscle controlling the pupil sizeControls the illumination on the retina3mm – 7mmIncrease by 5 timesCannot explain the 10 orders of magnitude light sensitivity

Slide 10 Aditi Majumder, UCI

Other

Pigmented EpitheliumBehind the retina – darkAbsorbs light and avoid scatteringAllows high-contrast sharp retinal image

Fovea2 degress of angle subtendedMost densely populated with photoreceptorsImage fixated on the retinaStiles-Crawford effect

Page 6: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

6

Receptors in Eye

Slide 12 Aditi Majumder, UCI

Receptor Organization

Page 7: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

7

Slide 13 Aditi Majumder, UCI

Receptors

Rods120 million Only in periphery

Cones6 million1% in fovea Rest in periphery

In periphery 20:1 ratio for rods and conesBlind Spot

Slide 14 Aditi Majumder, UCI

Receptor Distribution

Page 8: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

8

Slide 15 Aditi Majumder, UCI

Receptor Distribution

Turned away from lightTo get nutrition from opaque pigment epitheliumThe other cells are transparent not to block light reaching the retinaBlock the axons of the ganglion cells from leaving the eye

Slide 16 Aditi Majumder, UCI

Blind Spot

Ganglion nerve fibers fold and leave by crossing a part of retinaThis part does not have any receptorsFilled out by the brainExperiment

Page 9: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

9

Slide 17 Aditi Majumder, UCI

Receptor Functions

Cones – for vision in photopic conditionsRods – for vision in scotopic conditions

Saturate at high light levelsBoth – for vision in mesopic conditions

Slide 18 Aditi Majumder, UCI

Receptor Functions

Conversion of light to electrical energyLight sensitive chemical pigment in the receptorsAbsorbs photons and changes the shape to create a graded potential across the membrane of the outer segmentLogarithmic response supports Stevens LawPotential is transmitted down the outer membrane to other cells

Page 10: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

10

Slide 19 Aditi Majumder, UCI

Receptor Functions

Pigment BleachingPigment Regeneration

30 minutes for rods and 6 minutes for cone

Effects of Visual Pigments on Perception

Page 11: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

11

Slide 21 Aditi Majumder, UCI

Dark Adaptation Curve

Slide 22 Aditi Majumder, UCI

Dark Adaptation

Initial rapid increase 6 minutes

Slower further increase25-30 minutes

Rods have higher sensitivity than cones

Rods responsible for dark vision

Pigment Regeneration6 minutes for cones30 minutes for rods

Page 12: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

12

Slide 23 Aditi Majumder, UCI

Spectral Sensitivity

Threshold Curve Sensitivity Curve

Relative threshold of light required to detect monochromatic light of each wavelength

Slide 24 Aditi Majumder, UCI

Rod and Cone Sensitivity

Rods have sensitivity to shorter wavelengthsPeak near 500nm being maximum sensitive to the blue-green regionShift of wavelengths in dark

Purkinje shiftThings appear bluish in darkGreen foliage seems to stand out in the dusk

Page 13: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

13

Slide 25 Aditi Majumder, UCI

Cone Sensitivity

Cone curve is the combined sensitivity of three conesShort wavelength – S conesMedium wavelength – M conesLong wavelength – L cones

Slide 26 Aditi Majumder, UCI

Cone Sensitivity

Inconsistent with parts? – NORelative proportions of the S,M and L cones are differentS:M:L = 1:6:12Rods does not help in color vision. No color perception in dark

Page 14: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

14

Slide 27 Aditi Majumder, UCI

Why do we need three cones?

What happens with one cone?10 photons of 500nm produce x reaction10 photons of 560nm produce 4x reaction40 photons of 500nm produce 4x reaction

No way to tell between the wavelength

Slide 28 Aditi Majumder, UCI

Why do we need three cones?

What happens with one cone?Two parameters: Which wavelength and what brightness?Cannot measure both with one sensor

Page 15: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

15

Slide 29 Aditi Majumder, UCI

Why do we need three cones?

What happens with two cones?You can distinguish both intensity and wavelengthNeed more wideband sensitivity to cover all colorsMiss out completely parts of the spectrum

Slide 30 Aditi Majumder, UCI

Why do we need three cones?

With three conesWe do not need more than three since they cover the visible spectrum

Page 16: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

16

Slide 31 Aditi Majumder, UCI

Why rods not considered as the fourth cone?

Spatial distribution and nature of convergence are completely differentThey are activated usually in completely different light conditions

Effects of Neural Processing on Perception

Page 17: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

17

Slide 33 Aditi Majumder, UCI

Convergence

Measured number of neuron synapse on another neuron

Eye : Number of receptor neuron synapses on a ganglion cell

Each ganglion cell receives many synapsesRods to cones convergence ratio is 20:11 million ganglion cells, 120 million rods and 6 million conesRods have higher convergence than cones

Slide 34 Aditi Majumder, UCI

Rods are more sensitive in dark

Triggered

2 2 2 222

2

222

Threshold for ganglion triggering = 10

Not triggered

Rods have greater spatial summation than cones

Page 18: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

18

Slide 35 Aditi Majumder, UCI

Cones give visual acuity

Slide 36 Aditi Majumder, UCI

Cones give visual acuity

Page 19: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

19

Neural Circuits and their Effects on Perception

Slide 38 Aditi Majumder, UCI

Neural Circuits

Many neurons connected through convergenceSmall – a few neuronsLarge – a few hundred thousand neurons

Page 20: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

20

Slide 39 Aditi Majumder, UCI

Neural Processing by Excitation and Inhibition

Slide 40 Aditi Majumder, UCI

Neural Processing by Excitation and Inhibition

Page 21: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

21

Slide 41 Aditi Majumder, UCI

Neural Processing by Excitation and Inhibition

Slide 42 Aditi Majumder, UCI

Receptive Fields

The area of the retina that when stimulated influences the firing rate of the ganglion cellsExcitatory centerInhibitory surroundLateral Inhibition

Excitatory Center

Inhibitory Center

Neutral Region

Page 22: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

22

Slide 43 Aditi Majumder, UCI

Hermann Grid

Slide 44 Aditi Majumder, UCI

Explanation : Lateral Inhibition

Page 23: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

23

Slide 45 Aditi Majumder, UCI

Mach Bands

Actual Intensity

Percieved Intensity

Slide 46 Aditi Majumder, UCI

Explanation : Lateral Inhibition

Page 24: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

24

Slide 47 Aditi Majumder, UCI

Explanation : Lateral Inhibition

Slide 48 Aditi Majumder, UCI

Simultaneous Contrast

Page 25: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

25

Slide 49 Aditi Majumder, UCI

Simultaneous Contrast

Slide 50 Aditi Majumder, UCI

Explanation : Lateral Inhibition

Page 26: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

26

Artifact of Mach Band

Slide 52 Aditi Majumder, UCI

Theory of Continuity

Page 27: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

27

Slide 53 Aditi Majumder, UCI

Geometric Continuity

Slide 54 Aditi Majumder, UCI

Mach Bands

Artifacts when any patch or curve does not show C1 continuityThey are always avoided

ShadingBlending

Image editing, contrast compressing, tone reduction

Geometric Modeling

Page 28: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

28

Slide 55 Aditi Majumder, UCI

Shading

Id = kd (L . N)Is = ks (2N (L . N) – L) . V

Slide 56 Aditi Majumder, UCI

Shading

Flat Shading[I(v1)+I(v2)+I(v3)]/3Is not C1 continuous

Interpolated ShadingLinear Interpolation of I(v1), I(v2) and I(v3) Is C1 continuous

v1

v2

v3

Page 29: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

29

Slide 57 Aditi Majumder, UCI

Shading

Flat Shading – Mach Bands Interpolated Shading

Slide 58 Aditi Majumder, UCI

Intensity Blending

Proj1Proj2

Proj1Proj2

Overlap Region Overlap Region

Inte

nsity

0.0

Spatial Location

Depends onWidth of blendingBlending Function

Page 30: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

30

Slide 59 Aditi Majumder, UCI

Intensity Blending

Proj1Proj2

Proj1Proj2

Overlap Region Overlap Region

Inte

nsity

0.0

Spatial Location

Depends onWidth of blendingBlending Function

Slide 60 Aditi Majumder, UCI

Intensity Blending

Proj1Proj2

Proj1Proj2

Overlap Region Overlap Region

Inte

nsity

0.0

Spatial Location

Depends onWidth of blendingBlending Function

Page 31: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

31

Slide 61 Aditi Majumder, UCI

Intensity Blending

Slide 62 Aditi Majumder, UCI

Intensity Blending

Page 32: THE VISUAL SYSTEM - Donald Bren School of Information and Computer

32

Slide 63 Aditi Majumder, UCI

Image MosaicsMulti-Projector DisplaysImage Editing

Applications

Slide 64 Aditi Majumder, UCI

Geometric Modeling Primitives

Three propertiesContinuousContinuous with transformationsContinuous with subdivisions