RECTIFICATION

AND
RESTORATION
• Image rectification and restoration refer to
those operations that are preliminary to the
main analysis.
• It produces a corrected image that is as close
as possible, both radiometrically and
geometrically, to the radiant energy
characteristics of the original scene.
• In order to correct the image data, internal and
external errors must be determined.
• Internal errors are due to the sensor itself.
• These errors are systematic and constant in
nature, and can be known from pre-launch or
in-flight calibration measurements.
• External errors, which are unsystematic in
nature, are due to perturbations of the platform
and modulation of scene characteristics.
• These unsystematic errors can be determined
by relating known ground points to their
corresponding points in the image.
• Typical preprocessing operations include
radiometric and geometric corrections.
 Atmospheric correction
• While discussing the initial statistics, it was found
that remote sensing data in the visible bands are
affected by atmospheric effects due to scattering of
electromagnetic energy.
• This causes a degradation in the image quality.
• A black object tends to have some brightness, which
is an error.
• This error has to be removed.
• There are three methods by which this can be done:
1. Physical modeling
2. Histogram Minimum
3. Regression
 PHYSICAL MODELING
• These models permit observed reflectance to
be adjusted close to the true value that might
be observed under clear atmospheric
conditions.
• These models provide a rigorous and accurate
solution with flexibility to model a large variety
of atmospheric conditions.
• However, these also have many significant
disadvantages.
• Generally, these are complex models requiring
the use of intricate computer programs.
 HISTOGRAM MINIMUM METHOD (HMM)
• Atmospheric effects are generally low or negligible
in the infrared region and high in the visible region.
• When the histograms of the visible region are
examined, the lowest value of the histogram
amongst of the visible bands is subtracted from
the brightness values of all the pixels in the visible
bands.
• It is one of the simplest and direct method for
correcting atmospheric degradation, and is called
as Histogram Minimum Method (HMM).
• The advantages are its simplicity, directness, and
universal applicability, as it uses the information
present within the image itself.
• However, this method is an approximate one.
 REGRESSION METHOD
• The third category of correction method, while
retaining the concept of examining the brightness
of object within each scene, also accounts for the
interrelationship between separate bands.
• In this approach, a visible band is plotted against
an infrared band, and a best-fit (least-squares)
straight line is computed using standard
regression methods.
• The offset a on the x-axis as shown in Fig.
represents an estimate of the atmospheric
correction to be applied for that particular band.
• This method is known as the regression method.
The advantage of using regression method is that
it can be used for local areas also.
 Infrared Band

a
Visible Band

Regression method for computing atmospheric

correction
 GEOMETRIC CORRECTION
• All remote sensing imageries are inherently
subject to geometric distortions.
• These distortions may be due to several factors
such as:
(i) the perspective of the sensor optics,
(ii) the motion of the scanning system,
(iii) the motion of the platform,
(iv) the platform altitude, and velocity,
(v) the terrain relief, and
(vi) the curvature and rotation of the Earth.
 Geometric corrections
• Geometric corrections are intended to
compensate for those distortions which make
the geometric representation of the imagery
as close as possible to the real world.

• Many of these variations are systematic, or

predictable in nature, and can be accounted
for by accurate modeling of the sensor and
platform motion, and the geometric
relationship of the platform with the Earth.
• Other non-systematic or random errors cannot
be modeled and corrected in this way.

• Therefore, geometric registration of the

imagery to a known ground coordinate system
must be performed.

• The systematic and non-systematic errors in

remote sensing images are discussed below.
 SYSTEMATIC DISTORTIONS
• Scan skew: It is caused by the forward
motion of the platform during the time required
for each mirror sweep. The ground swath is
not normal to the ground track but is slightly
skewed, producing cross-scan geometric
distortion.

SCAN SCREW
• Mirror scan velocity: The MSS mirror
scanning rate is usually not constant across a
given scan, producing along-scan geometric
distortion.
EL GNA ACTUAL
R ORRI M

NOMINAL

TIME
MIRROR VELOCITY VARIATION
• Panoramic distortion: The ground area
imaged is proportional to the tangent of the
scan angle rather than to the angle itself.
Since data are sampled at regular intervals,
this produces along-scan distortion.

YAW VARIATION
• Platform velocity: If the speed of the platform
changes the ground track covered by
successive mirror scans changes producing
along-track scale distortion.

SPACECRAFT VELOCITY
• Earth rotation: The Earth rotates as the MSS
sensor scans the terrain.

• This results in a shift of the ground swath

being scanned, causing along-scan distortion.

EARTH ROTATION
• Perspective: For some applications it is
desirable to have images represent the
projection of points on the Earth upon a plane
tangent to the Earth with all projection lines
normal to the plane. This introduces
along-scan distortion.
 NONSYSTEMATIC DISTORTIONS
• Altitude: If the platform departs from its normal
altitude, changes in scale occur.

• Attitude: One of the sensor system axes usually

maintained normal to the earth’s surface and
introduces geometric distortion.

ALTITUDE VARIATION
ROLL VARIATION
 GEOMETRIC RECTIFICATION

• The basic purpose of carrying out geometric

correction is to orient the original data collected so
that it has the desired scale and map projection or
simply, it is made planimetric.
• A map projection is a device for the representation
of a curved surface on a flat sheet of paper.
• Different types map projections have been
developed, however, each projection tries to
preserve only some property of the mapped area,
and either it is uniform representation of areas or
shapes, or preservation of correct bearings.
• Only one such property can be correctly
represented, though several projection attempt to
compromise by minimizing distortion in two or more
map characteristics.
• Further, information extracted from remote sensing
data may be used with other geographical maps, or
two adjacent images of an area or multi temporal
images are to be used for assessing the changes
that may have occurred, the satellite data has to be
fitted to a coordinate system.
• This process of providing a coordinate system to an
image is known as registration.
• Thus, geometric correction of remotely-sensed
images required when the image is to be used in
one of the following circumstances
(i) to transform an image to match a map
projection,
(ii) to locate points of interest both on map and
image,
(iii) to bring adjacent images into registration,
(iv) to overlay temporal sequences of images of the
same area, and
(v) to overlay images and maps within a GIS.
• Another factor which causes error is the motion of
the earth as the data is being collected by the
sensor.
• In case of Landsat TM sensor, there are 16 sets of
sensors i.e. data for 16 scan lines are acquired at a
given interval of time ‘t’.
• The next 16 scan lines i.e. scan lines no. 17-32 will
be acquired at time‘t+1’.
• However, during the time period of t and t+1, the
earth has moved east ward.
• So the scan line 17-32 will start slightly west ward
than scan line 1-16 and also end slightly before the
east ward limit of the previous 16 scan lines.
• This shift causes a skew in the image coverage.
• The net effect of this is shown in Figure.
 A O B

M scan lines

D’ D Q P C’ C

• It may be noted, that as the angular velocity of Earth

varies with latitude, so this skew becomes more
prominent at higher latitudes.
• So, the geometric correction process can be
considered to include:-

i. Determination of a relationship between the

coordinate system of map and image (or image and
image in the case of registration),
ii. Establishment of a set of points defining pixel
centers in the corrected image with that, when
considered as a rectangular grid, define an image
with the desired cartographic properties, and
iii.Estimation of pixel values to be associated with
those points.
• The relationship between the two coordinate
systems (map and image) could be defined by two
methods.

i. The orbital geometry of the satellite platform.

ii. The transformation using ground control points.
 ORBITAL GEOMETRY MODEL
• Orbital geometry methods are based on knowledge
of the characteristics of the orbit of the satellite
platform.
• It is a simple method of correcting the coordinate
system of remotely-sensed images using
approximate orbit parameters.
• This method is suitable only for use for images that
have been acquired from sensor having narrow
angular field of view, such as the Landsat TM and
the SPOT HRV.
• However, it can be easily extended to deal with
imagery from any similar sensor, provided that the
nominal orbital parameters of the satellite are known
or can be estimated.
• Since the method is based upon nominal orbital
parameters, the accuracy of the geometrically
corrected image produced by this technique is not
high.
• In the orbital geometry model, the correction can be
modeled in three sequential steps as enumerated
below:

(i) Aspect Ratio

(ii) Skew Correction
(iii) Earth rotation correction
 ASPECT RATIO
• Every sensor as per it Instantaneous Field of
view, has a ground coverage i.e. its pixel size.
• As the sensor moves along a scan line, there is
a small overlap, so instead of a square pixel,
now it is a rectangular pixel.
• In case of Landsat MSS, the actual size of the
collected data is 79m×56m.
• Thus, the shape of pixel is to be restored back to
it desired square shape.
• This correction is known as the aspect ratio
correction and can be represented by the
transformation matrix, M1
1.41 0.00
M1 =  
 0. 00 1. 00 
 SKEW CORRECTION
• Low altitude satellites are general placed in a
near North-South axis.
• However, to counter the forces due to flattening
of the Earth, precessing of the earth axis and the
forces due to stars, it is not possible to place the
satellite in a perfect North-south alignment.
• Landsat satellite 1 to 3 has an orbital inclination
of 99.09o, whereas Landsat 4 and 5, have an
inclination of 98.2o.
• The satellite heading (across-scan direction at the
Equator) is therefore 9.09 o and 8.2o respectively,
increasing with latitude.
• The skew angle θ at latitude L is given (in degrees)
−1  sin θ E 
by
θ = 90 − cos  
 cos L 

where is the satellite heading at the Equator

and L is the centre latitude of the image.
• Given the value of the coordinate system of the
image can be rotated through degrees anticlockwise
so that the scan lines of the corrected image are
oriented in an east-west direction using the
transformation matrix M2:
 cos θ sin θ 
M2 = 
− sin θ cos θ 
 EARTH ROTATION CORRECTION

• As the satellite moves southwards along its orbit,

the Earth rotates beneath it in an easterly direction
with a surface velocity proportional to the nadir
point.
• To compute the displacement of the last line in the
image relative to the first line, determine the time
taken by the satellite sensor to scan the image and
the eastwards velocity of the Earth.
• The time taken for the satellite sensor to scan the
image can be found if the distance traveled by the
satellite and the satellite’s velocity are known.

• Both distance and velocity are expressed in terms of

angular measure.
• If A is a point on the Earth’s surface corresponding
to the centre of the first scan line in the image, and
B is the corresponding point on the last scan line in
the image, then the line AB represents an arc of a
circle centered at the Earth’s centre.
• The angle at the Earth’s centre O is given by angle
AOB and can be calculated as the Earth’s equatorial
radius (R) is 6378 km, so
Angle AOB = Arc (AB/R)
N

W θ E

Equator

Direction of
Satellite
movement

S
• Further, the orbital period (P) of the satellite is
known, thus the angular velocity of the satellite (ω)
is equal to 2π/(P x 60).
• The time (t) required to travel the angular distance
AOB can be obtained by dividing ω/Angle AOB.
• Subsequently, determine the displacement of the
last scan with respect to first scan line.
• This is dependent on the central latitude (L) of the
image.
• If the Earth’s surface velocity at latitude L is
VE(L), it can be defined as:
V E ( L ) = R cos( L )ω E
where R is the Earth’s radius, defined above,
and is the Earth’s angular velocity
• Since the Earth rotates once in 23 hours, 56
minutes and 4 seconds (i.e 86164 seconds), then its
angular velocity is simply 2π/86164 rad/sec or
0.7279 rad/sec.
• Now at the given Latitude (L) determine the skew
angle θ.
• So actual displacement is ω 0 × t × cos θ .
• These computations are summarized by the term a sk
ω e cos L cos L
a sk = = 0.0719
ω 0 cos θ cos θ

where is the Earth’s angular velocity ωe, is the

satellite’s angular velocity ω0(both given above),
and θ and L are defined above.
The transformation
M 3 matrix M3 is:
1 a sk 
= 
 0 1 
• The three transformation matrices M1, M2 and M3
given above are not applied separately.
• Instead, a composite transformation matrix, M, is
obtained by multiplying the three separate
transformation matrices:
M = M 1M2M3
• The corrected image coordinate system is related to
the raw image coordinate system by
x’= Mx
where is x’=(x’1,x’2) the vector defining the
coordinates of the corrected pixel and x=(x 1,x2) is
the original coordinate.
 GEOMETRIC RECTIFICATION
• Geometric rectification is the process by which
the geometry of an image is made planimetric.

• It requires the use of GCPs whose image

coordinates in terms of rows and pixel number
are known along with the map or ground
coordinates.
• It is an accurate process since each pixel is
referenced not only by its image coordinates
but also by a ground coordinate system in a
standard map projection, and hence known as
image to map rectification.
• Geometric rectification has to be done when
accurate linear or areal measurements are
needed.
• Many a times accurate geometric rectification
is not needed especially when two or more
images are used to find out the changes in an
area or region.
• Here instead of carrying geometric
rectification of all the images and then
determining the changes, the images are
registered with respect to each other.
• This is known as image registration or
image-to-image registration.
 GEOMETRIC CORRECTION
• To remove systematic errors, it requires an
in-depth knowledge of a large number of
parameters related to satellite ephemeris data
and these errors are corrected at the Master
Data Centre.
• Generally, non-systematic errors are corrected
by the user himself by selecting suitable
number of GCPs.
• There are two geometric correction
procedures, i.e., geometric rectification and
geometric registration, which are generally
applied.
• The difference between rectification and
registration is that in rectification the image is
referenced to a map having a standard map
projection while in registration two images are
referenced to each other and not to any map
having a standard map projection.

• As most of the remote sensing applications yield

a map, hence image to map rectification has
been discussed in details.
RECTIFICATION
 MAP RECIFICATION
• In order to rectify a remote sensing image to a
map coordinate system, two basic operations
have to be performed

(i) spatial interpolation

(ii) intensity interpolation.
• In spatial interpolation, a large number of GCPs are
identified on the image and ground.
• A polynomial is fit to the control points using least
squares criteria to model the geometric error directly
into the image without explicitly identifying the
source.
• Generally, a six-parameter affine transformation is
sufficient to rectify a remote sensing image as it can
model for six kinds of distortions, i.e., translation in x
and y, scale changes in x and y direction, skew, and
rotation.
• This can be expressed mathematically by the
following equations:

x’ = a0 + a1x + a2y
y’ = b0 + b1x + b2y

where x and y are the ground coordinate

values and x’ y’ are the image coordinates.

• It may be noted that minimum of 3 GCPs can

yield a solution, however more than 3 GCPs
are used to computer the coefficients a0, a1, a2,
bo, b1, and b2.
• This method involves the computation of root
mean square error (RMSerror) for each GCP.

RMSerror = ( x' − x0 )2 + ( y' − y0 )2

where (x0,y0) and (x', y') are the original and
computed image coordinates of the GCP
respectively.
• By computing RMSerror for all the GCPs, it is
possible to identify the GCP which exhibits the
greatest error and also the sum total of RMSerror of
all GCPs.
• Generally, the analyst specifies a threshold
value for total error.
• If the total error exceeds the threshold value,
then the GCP having the highest individual
RMSerror is deleted, and a new set of
coefficients of ao, a1, a2, bo, b1, and b2 are
computed.
• This process is an iterative in nature and
continues till total RMSerror of all the GCPs is
within a user specified threshold, or the
number of GCPs falls below an acceptable
limit, usually 3.
• Once the acceptable RMSerror is achieved, the
intensity interpolation of geometrically rectified
image starts.
 INTENSITY INTERPOLATION

• After the spatial interpolation is carried out, the

next step is to extract or identify the brightness
values of new image from the old image.
• This process is known as resampling.
• The resampling process calculates the new
pixel values from the original digital pixel values
in the uncorrected image.
• There are three common methods for resampling:-
(i) nearest neighbour,
(ii) bilinear interpolation, and
(iii) cubic convolution.
 Nearest Neighbour
• Nearest neighbour resampling uses the digital
value from the pixel in the original image which
is nearest to the new pixel location in the
corrected image.
• This is one of the simplest methods and does
not alter the original values, but may result in
some pixel values being duplicated while
others are lost.
• This method also tends to result in a disjointed
or blocky image appearance.
INTENSITY INTERPOLATION
 BILINEAR INTERPOLATION
• Bilinear interpolation resampling takes a
weighted average of four pixels (BVwt) in the
original image nearest to the new pixel
location.

∑(z )
4

k / Dk2
BVwt = k =1

∑ (1 / D )
4
2
k
k =1

• The average process alters the original pixel

values, and creates entirely new digital values
in the output image.
 CUBIC CONVOLUTION
• Cubic convolution resampling goes even
further to calculate a distance weighted
average of a block of sixteen pixels from the
original image which surrounds the new output
pixel location.
• As in bilinear interpolation, this method also
results in a completely new pixel value.
• However, both the methods produce images
which are sharper in appearance and avoid
the blocky appearance of the nearest
neighbour method.
 Method Technique Relative Advantages Disadvantages
computing

Nearest Transfers the 1 Simple to Image

neighbors brightness value of compute array disjointed
the nearest pixel. values or blocky
unaltered
Bilinear Proximity 10 Smooth image, Grey
weighted average geornetrically values
method from accurate altered.
nearest pixels
Cubic Proximity 20 Very smooth Complex
Convolution weighted average image to
method of sixteen compute,
nearest pixels. gray value
altered.