• Raw remotely sensed image data contain flaws

or deficiencies and correction is required prior to

image processing.

Image Pre-Processing

• Raw remotely sensed image data contain flaws Geometric correction

or deficiencies and correction is required prior to
image processing.
• Geometric correction is the transformation of a
remotely sensed image such that it has the scale
• This correction is termed image pre-processing. and projection properties of a map.
Depending on the source of error, deficiency
correction and flaw removal are divided into two
categories:
– radiometric and
– geometric correction.

Geometric correction
• The most common geometric correction method
• Geometric correction is the transformation of a
remotely sensed image such that it has the scale
and projection properties of a map.
• Geometric errors are due to:
– instrument error
– the curvature of the earth
– earth rotation during image acquisition and
– platform instability
is based on ground control points (GCP). A
number of GCPs are defined on each of the
images you want to correct. The best GCPs are:
– road intersections,
– airport runways,
– edges of dams or buildings,
– corners of agricultural fields
– other permanent features.

1
• GCPs should be:
– sufficient in number and
– evenly spread across the scene.

• Image to Ground Geocorrection

The correction of digital images to ground coordinates using ground
control points collected from maps or ground GPS.

Ground control points (GCPs): locations that are found both in the
image and in the map. The GCP coordinates are recorded from the
map and from the image and then used in the least squares
regression analysis to determine the coordinate transformation
equation

• Image to Image Geocorrection

Image to Image correction involves matching the coordinate
systems of two digital images with one image acting as a reference
image and the other as the image to be rectified.

Ground Control
Points

Radiometric correction

• Radiometric errors can be due to:

– data recording instrumentation (e.g. incorrect

calibration of detectors or incorrect function)

– atmospheric noise that result in the signal recorded

by the sensor not representing the signal reflected by
the target.

2
 Sensor Calibration

• Radiometric correction includes:

• Sensor calibration is important when:

– applying sensor calibration

– images from different sensors are used
– replacing missing scan lines
– time-series images are used for pixel value
– de-striping comparison and
– applying atmospheric correction – parameters from the images such as reflectance and
temperature need to be extracted.

• Raw data are converted from digital numbers (DN) to

Table 11.2 ETM+ Spectral Radiance Range
units of radiance using the following equation: watts/(meter squared * ster * µm)

Before July 1, 2000 After July 1, 2000

⎛ ( L max − L min) ⎞
Radiance = ⎜ ∗ QCAL ⎟ + L min Low Gain High Gain Low Gain High Gain
⎝ QCAL max ⎠
Band LMIN LMAX LMIN LMAX LMIN LMAX LMIN LMAX
Where: Number
1 -6.2 297.5 -6.2 194.3 -6.2 293.7 -6.2 191.6
Lmin = spectral radiance at the minimum quantized and
calibrated data digital number 2 -6.0 303.4 -6.0 202.4 -6.4 300.9 -6.4 196.5

3 -4.5 235.5 -4.5 158.6 -5.0 234.4 -5.0 152.9

L max = spectral radiance at the maximum quantized and
calibrated data digital number 4 -4.5 235.0 -4.5 157.5 -5.1 241.1 -5.1 157.4

5 -1.0 47.70 -1.0 31.76 -1.0 47.57 -1.0 31.06

QCAL max = 255
6 0.0 17.04 3.2 12.65 0.0 17.04 3.2 12.65

QCAL = DN 7 -0.35 16.60 -0.35 10.932 -0.35 16.54 -0.35 10.80

8 -5.0 244.00 -5.0 158.40 -4.7 243.1 -4.7 158.3

• A reduction in between-scene variability is achieved

through normalization for solar irradiance by converting
spectral radiance to reflectance using the following
equation:
(π ∗ Lts ) d2 1 Astronomical Unit = nearly 150 million kilometres
ρ ts =
E0 *cos(θ 0 ) *ESUN
d λ
Where:

ρts = reflectance at the top of the atmosphere

Lts = spectral radiance at the sensor’s aperture

d = Earth-Sun distance interpolated from Table (next page-based
on Julian day) in Astronomical Units

ESUNλ = solar irradiance at the top of the atmosphere

θ 0 = solar zenith angle

3
 • Earth's orbit is elliptical, which means it's
closest to the Sun once a year (the first
week in January), and farthest from the
• Julian day = The number of days elapsed Sun once a year (in early July).
since the start of the year
• Early July = Julian Day?
31 days (January)
+
28 days (February)
31 days (March)
30 days (April)
31 days (May)
30 days (June)

= 181 Julian Days

Table 11.3 ETM+ Solar Spectral Irradiances

d
Table 11.4 Earth-Sun Distance in Astronomical Units
Band ESUNλ (Watts x m-2 x µm-1) (Watts x cm-2 x µm-1)
Julian Julian Julian Julian Julian
Dist. Dist. Dist. Dist. Dist. 1 1969.000 0.1969
Day Day Day Day Day

2 1840.000 0.184
1 .9832 74 .9945 152 1.0140 227 1.0128 305 .9925
3 1551.000 0.1551

15 .9836 91 .9993 166 1.0158 242 1.0092 319 .9892 4 1044.000 0.1044

5 225.700 0.02257
32 .9853 106 1.0033 182 1.0167 258 1.0057 335 .9860
7 82.07 0.008207
46 .9878 121 1.0076 196 1.0165 274 1.0011 349 .9843
8 1368.000 0.1368

60 .9909 135 1.0109 213 1.0149 288 .9972 365 .9833

Input image

$n2_

Equations

((Lmaxband1 – Lminband1)/255)* ($n1_)) + Lminband1

Output image

4
• The thermal band of Landsat 7 ETM+ (band 6) is converted from
spectral radiance to temperature. The conversion equation is:

K2 • Radiometric correction includes:

T=
K1
ln( + 1)
Lλ – applying sensor calibration
– replacing missing scan lines
Where:
– de-striping
T = effective at-satellite temperature in Kelvin
– applying atmospheric correction
K 1 = calibration constant = 666.09 ( W * m −2 * ster −1µ m −1 )

K 2 = calibration constant = 1282.71 (o Kelvin)

Lλ = spectral radiance ( W * m −2 * ster −1µ m −1)

• Replacing Scan lines • Replacing Scan lines

• Missing scan lines occur when a detector fails to operate • Missing scan lines occur when a detector fails to operate
during a scan. This, results in a zero brightness value in during a scan. This, results in a zero brightness value in
each pixel of the particular line and will appear black in each pixel of the particular line and will appear black in
the image. the image.

• There is no way to restore these values but it is possible

to improve the visual interpretation of the data based on
spatial autocorrelation. This relies on the fact that points
close in geographical space tend to have similar values
of the variable of interest. This is the basis for the
estimation of missing brightness values in scan lines.

Effect of a missing scan line

• Replacing Scan lines
Zero brightness value appears black on the image

• Missing scan lines occur when a detector fails to operate

during a scan. This, results in a zero brightness value in
each pixel of the particular line and will appear black in
the image.

• There is no way to restore these values but it is possible

to improve the visual interpretation of the data based on
spatial autocorrelation. This relies on the fact that points
close in geographical space tend to have similar values
of the variable of interest. This is the basis for the
estimation of missing brightness values in scan lines.

• The average brightness value between the pixel in the

preceding and the succeeding scan lines is then
assigned to the missing pixel value.

5

• Radiometric correction includes:
– applying sensor calibration
– replacing missing scan lines
– de-striping
– applying atmospheric correction
• Radiometric correction includes:
– applying sensor calibration
– replacing missing scan lines
– de-striping
– applying atmospheric correction
Scattering is caused by atmospheric gases
such as oxygen, nitrogen, ozone and other
particles.
The effect of scattering is the redirection of
electromagnetic radiation.
Absorption is caused by water vapour, carbon
dioxide and ozone.
De-stripping
• Line striping is caused when detectors are
imbalanced.
Atmospheric Correction
• Atmospheric constituents such as:
– gases,
– aerosols (airborne particulate matter) and
– clouds,
scatter and absorb:
(a) solar radiation travelling through atmosphere
before reaching the earth and
(b) reflected electromagnetic radiation from the earth
surface before detection by the satellite sensor.
These two processes result in different recorded
values of electromagnetic radiation at a given location
than at the respective ground location.
However, in the electromagnetic spectrum there are
atmospheric windows where the effects of absorption
and scattering are minimised.
Remote sensing is concentrated in bands within the
atmospheric windows which are least affected by
absorbing molecules.
6
 • The most common atmospheric correction
• Atmospheric Correction is important when:
method is the darkest pixel (DP) method.

• Comparing images with ground data

• DP is the most widely used method and is
image based. It does not require any other
• More than one image is used since information apart from that contained in
satellite data are likely to be recorded image.
under different atmospheric conditions.

• The visible spectrum is affected by atmospheric

scattering whilst the near infrared is free from
scattering.

• Water has very low reflectance values in the

visible bands and negligible in the near infrared
band. The DP method assumes that the pixel
with the lowest digital number in each band
should be zero and its actual value is due to
atmospheric effects.

• The pixel with the lowest number may

correspond to a water body or other dark
object in the image.