GEOM30009: Imaging the Environment
Human vision and principles of image
formation
Kourosh Khoshelham
Topics
• Human vision: how we see
– Perceiving light
– Perceiving colour
• Image formation
– Sensing light
– Image structure
– Image resolution
– Colour theory
2
Human vision
• Image formation and
imaging sensors are based
on the way human eye
works;
• Light passes through
cornea, pupil and the lens;
• It is then sensed by the
cells on the retina;
3
Perceiving light
• The iris can make the pupil
larger or smaller allowing
more light or less light into
the eye;
• Retina contains a chemical
that produces electrical
impulses based on the
amount of received light;
• The brain processes the
impulses into what we see.
4
Perceiving light
• The retina contains two types of cells:
– Rod cells sense grey levels and operate
at low light;
– Cone cells are responsible for colour
vision and detail.
• The fovea region has a concentration of
cone cells and is responsible for seeing
fine details.
5
Perceiving colour
• The cone cells produce a
chemical that contains colour-
sensitive pigments;
• Three types of pigments:
– Sensitive to red colour;
– Sensitive to green colour;
– Sensitive to blue colour.
• Each cone cell has one of the
three types of pigments and is
sensitive to one colour.
6
Image formation by imaging sensors
Most imaging sensors are
very similar to the eye:
Filter Cornea
Aperture Iris
Camera
lens
Eye lens
Film/Sensor
plate
Retina
Sensing
material
Rods/Cons
7
Sensing light
• A sensor is often an array
of photodetectors;
• A photodetector produces
an output voltage that
depends on the amount of
received light
(electromagnetic energy)
and the sensing material.
• The filter allows only a
certain range of
wavelengths to reach the
sensor.
8
Digital image acquisition
• An A/D converter
converts the electric
charge to a digital
number;
• The digitized image is
then stored in a matrix
structure (aka raster
format);
• Each element of the
matrix is called a pixel.
9
Image resolution
• The size of each pixel determines the spatial resolution of
the image;
• The number of bits used to store the digital number for
each pixel determines the radiometric resolution.
L = 2k
number of
grey levels
number of bits
Bits Grey levels
1 2
2 4
…
8 256
16 65536
10
Image resolution
Spatial resolution defines the ability to resolve spatially close objects and
to identify small objects.
Expressed by Ground Sampling Distance (pixel size on the ground).
GSD depends on image scale and the pixel size.
0.80m 0.40m 0.20m 0.10m 0.01m
11
Image resolution
Radiometric resolution defines the ability to resolve objects with
similar reflectance (within the same wavelength).
8-bit image1-bit image
12
Mosque of Hassan II, Casablanca, Morocco
Image resolution
• In imaging the environment there are two additional types of
resolution that are also important:
– Spectral resolution: the ability to resolve objects with similar
spectral features;
→ Narrower spectral bands = Better spectral resolution
– Temporal resolution: the ability to resolve events or phenomena
that occur in a short time interval;
→ Shorter revisit/reacquisition cycle = Better temporal resolution
13
Image resolution
Example: improved spectral resolution of Landsat 8 instrument (OLI)
Landsat 4-5 (MSS) Wavelength (nanometres) bandwidth (nanometres)
Band 1 500-600 100
Band 2 600-700 100
Landsat 8 (OLI) Wavelength (nanometres) bandwidth (nanometres)
Band 1 430-450 20
Band 2 450-510 60
Band 3 530-590 60
Band 4 640-670 30
14
Multispectral imaging
• Multispectral sensors can capture images at multiple wavelengths of
the electromagnetic spectrum.
• This is typically done by using a set of beam splitters and filters
which separate different wavelengths and direct them to different
photodetectors.
D
D
D
D
D Visible detectors (8)
Spectral balance filter
Dichroic beamsplitt er (Db)
Db
Collimating lens
Mirror
Primary mirror
Secondary mirror
Mirror
Mirror
Db
focus
lens
vacuum
window
cooler
window
10.7 m
12 m
6.75 m
3.9 m
Collimating
lens
filter
filter
filter
Depolarizer
Visible bandp ass filter
Radiant flux ()
from the terrain
4
5
3
2
1
filter
0.52 – 0.72 m
Example:
GOES
multispectral
imager
15
Multispectral imaging
A multispectral image set comprises multiple images each captured
at a certain wavelength band.
Seven Bands of Landsat
Thematic Mapper Data of
Charleston, SC, Obtained on
February 3, 1994.
Source: Jensen, 2000.
16
Colour image formation
• A colour image is formed by displaying three images captured at
different wavelengths using the red, green, blue primary colours.
17
Red
Green
Blue
RGB
Colour theory
• Colour theory:
– Additive colour
– Subtractive colour
Additive colour
Example: mixing different wavelengths of
light in a computer monitor. White light is
the combination of all wavelengths in the
visible range.
Subtractive colour
Example: mixing paint in a colour printer.
Black is the combination of different
paints that absorb different wavelengths
of visible light.
18
Image display using additive colour
• Image display devices are based on additive colour.
• Primary colours: Red, Green, Blue (RGB).
• Mimics the operation of the eye.
19
True colour vs false colour image formation
20
True colour False colour (color-infrared)
R-G-B NIR-R-G
Red
channel
Green
channel
Blue
channel
400nm 500nm 700nm 900nm …
Red
channel
Green
channel
Blue
channel
400nm 500nm 700nm 900nm …
True (natural) colour image
• R: red band image
• G: green band image
• B: blue band image
red band image
Green band image
Blue band image
21
False colour image
• R: spectral band i
• G: spectral band j
• B: spectral band k
SPOT 5
band 3
(infrared)
SPOT 5
band 2
(red)
SPOT 5
band 1
(green)
SPOT-5 sample image of Naples (Italy) in 2002 (image credit: CNES)
22
Grayscale image
• R: spectral band i
• G: spectral band i
• B: spectral band i
SPOT-5 sample image of Naples (Italy) in 2002 (image credit: CNES)
SPOT 5
band 3
(infrared)
23
Pseudo-colour image
• In pseudo-colour images, colour is used to show a certain property of
the surface such as height, temperature, or vegetation density.
• A single image → a colourmap → R, G, B
Height R G B
-10 0.00 0.00 0.50
-8 0.00 0.00 1.00
-6 0.00 0.25 1.00
-4 0.00 0.50 1.00
-2 0.00 1.00 0.00
0 1.00 1.00 0.00
2 1.00 0.75 0.00
4 1.00 0.00 0.00
6 0.75 0.50 0.50
8 1.00 1.00 1.00
An example colourmap:
Elevation model of Mars
24
Quantitative colour description
Based on the intensity of primary colours represented by a value between 0 and 1.
Red Green Blue Colour
0 0 0
0 0 1.00
0 0.25 1.00
0 0.50 1.00
0 0.75 1.00
0 1.00 1.00
0.25 1.00 0.75
0.50 1.00 0.50
0.75 1.00 0.25
1.00 1.00 1.00
1.00 1.00 0
1.00 0.75 0
1.00 0.50 0
1.00 0.25 0
1.00 0 0
Summary
• Anatomy of the eye
– Perceiving light and colour
– Image formation in the eye
• Image acquisition by digital sensors
– Image structure;
– Image resolution → spatial, temporal, radiometric, spectral
– Multispectral imaging
• Colour theory and image formation
– True colour, false colour, grayscale, pseudo-colour
26
© Copyright The University of Melbourne 2011