Title: Human Vision
1Human Vision
- Its all in your head.
- -- Kristen Pelz 06, Biology Nerd
2The Human Eye
3Cornea
- Filled with aqueous fluid that provides nutrition
to the surrounding tissue - Fixed, dome shape refracts light towards the the
iris - Ensures that light converges to the lens at the
pupil
4Iris
- Colored ring of muscle fibers
- By contracting and expanding, the iris controls
the amount of light that enters the pupil
5Lens
- Lens has natural elasticity
- Ciliary muscle stretches lens, elasticity
retracts lens - Presbyopia (farsightedness) is a problem during
aging because the elasticity of the lens wears out
6Optics
- The cornea converges light on the lens
- The lens refracts and focuses light on the retina
7Retina
- Parts
- Fovea
- Blind spot
- Rods
- Cones
- 130 million photoreceptors
- Covers 72 of the sphere of the eye
- Responds chemically to photo stimulus
- Does some pre-processing for the brain
8Fovea
- The fovea is a dense, hexagonally-packed
depression of cone-type photoreceptors at the
back of the eye - Only comprises 2 of the field of view, but
contains 20 of the eyes photoreceptors
(reference lost) - Vital for central vision
- Saccades place the fovea on an object of interest
9Rods
- 125 million rods in the eye
- Active at low light-levels
- Black and white (only respond to intensity)
- Several rods connect to one optic nerve cell
(ganglion cell) to elicit even more
light-sensitivity.
10Cones
6 million cones Concentrated in the fovea in a
tight hexagonal pattern Active only at high
light-levels (when rods have saturated) Three
different types of cones respond to three
different colors of light red, green, and blue.
One cone attaches to a single ganglion cell
11How does it come together?
- Cones, concentrated in the fovea, give humans
high-resolution color vision in the very center
of the field of view. - Rods, on the periphery and scattered through the
retina, give us night vision, though not in
color. - When a rod or cone is stimulated by light, it
releases a photopigment that is received by a
ganglion cell. If there is enough pigment, the
ganglion cell fires an action potential into the
vision pathway along the optic nerve - In high light-levels, the photopigment bleaches
out, lowering the effective brightness. This is
why it can take up to thirty minutes for ones
eyes to adjust to indoor lighting after being
exposed to the bright sun. - The ganglion cells have receptive fields - tiny
regions in the field of view that are seen by
their attached photoreceptors (rods and cones).
Using these receptive fields, the retina is able
to do a great deal of preprocessing condensing
the information from 130 million photoreceptors
down to a signal on only 1.2 million nerve axons.
12The Visual Pathway
- Optic Nerve
- 1.2 million fibers
- Causes a blind-spot in the retina
- Transports information from the retina to the
optic Chiasm
13Optic Chiasm
- The axons from the left half of the right eye
cross over the axons from the right half of the
left eye - The right half of the brain handles the left half
of the field of vision, not the left eye - The left half of the brain handles the the right
half of the field of vision
14Lateral Geniculate Nucleus
- In the thalamus
- Optic tract is input from optic chiasm
- Optic radiations are outputs to the brain
- The LGN combines information from other senses to
make predictions - Two types of cells M cells (layers 1 and 2) and
P cells (layers 3, 4, 5 and 6) - M cells process data quickly, with no color
information, and ask the question where? - P cells process data more slowly, use color
information, and ask what? -
15Visual Cortex
- Approximately 1/3 of the neo-cortex (the part of
the brain most responsible for intelligence) - Columns of neurons perform input-based
learning, and specialize into different roles - Four main regions
- Primary visual cortex (V1)
- V2
- V3
- V5 (also know as MT, or middle/medial temporal)
16Primary Visual Cortex - V1
- Spatially equivalent to the field of view
- Low-level in the computer vision hierarchy
input of a processed image, output of simple
geometric data (edges, connections, etc.) - Detects fine details
- Visual Orientation
- Spatial Frequency
- Colors
- Motion
- Direction
- Speed
17V2
- Feed-forward from V1
- Processes figure/ground separation
- Orientation of illusory contours
- Mid-level in the computer vision hierarchy
input is highly processed image data and output
is simple scene information - Strong feedback to V1 (for prediction?)
18V3 and V5
- V3 processes global motion
- V5 processes the movement of complex objects and
performs high-level object recognition - V3 and V5 both fall into high-level vision in
the computer vision hierarchy, though V5 is
definitely more abstract. Using V5 is thinking
visually.
19The End
Peter Lubans March 2006 For more information,
visit www.wikipedia.org thats where I did all
of the research and obtained all of the images
for this presentation
20Sources
Optic radiations. Wikimedia, August 2005.
http//en.wikipedia.org/wiki/Optic_radiations
Last accessed March 2006. Eye. Wikimedia,
March 2006. http//en.wikipedia.org/wiki/Eye
Last accessed March 2006 Optic Nerve.
Wikimedia, Frebruary 2006. http//en.wikipedia.or
g/wiki/Optic_nerve Last accessed March
2006 Optic Chiasm. Wikimedia, February 2006.
http//en.wikipedia.org/wiki/Optic_chiasm Last
accessed March 2006 Optic Tract. Wikimedia,
January 2006. http//en.wikipedia.org/wiki/Optic_
tract Last accessed March 2006 Retina.
Wikimedia, March 2006. http//en.wikipedia.org/wi
ki/Retina Lateral Geniculate Nucleus.
Wikimedia, February 2006. http//en.wikipedia.org/
wiki/Lateral_geniculate_nucleus Last accessed
March 2006. Visual Cortex. Wikimedia, February
2006. http//en.wikipedia.org/wiki/Visual_cortex
Last accessed March 2006. Jeff Hawkins, On
Intelligence. Times Books, New York, New York,
2004.
21Images