University of Technology

Building & Construction Department / Remote Sensing & GIS lecture

5. Corrections
5.1 Introduction
5.2 Radiometric Correction
5.3 Geometric corrections
5.3.1 Systematic distortions
5.3.2 Nonsystematic distortions
5.4 Image Rectification
5.5 Ground Control Points (GCPs)
5.5.1 Selection and minimum number of GCPs
5.6 Geometric Correction Models (transformations)
5.6.1 Polynomial Transformations
5.7 Root Mean Square Error (RMSE)
5.8 Resampling methods
5.8.1 Nearest Neighbor
5.8.2 Bilinear Interpolation
5.8.3 Cubic Convolution

| P a g e ( 1 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

5.1Introduction

Remote sensing system gathered data in many forms and techniques. In all
these systems, there are numerous errors associated with gathered data. Therefore,
it is usually necessary to preprocess the remotely sensed data in order to remove
these errors. Image preprocessing operations normally precede all other image
manipulation and analysis, such as enhancement or classification. The
preprocessing of remotely sensed image (image restoration) consists of geometric
and radiometric characteristics analysis. by realizing these features, it is possible to
correct image distortion and improve the image quality and readability. Image
restoration is usually implemented to correct distortion, to remove degradation, and
to reduce noise introduced during the image process. Image restoration produces a
corrected image that is as close as possible, both geometrically and radiometrically,
to the radiant energy characteristics of the original scene. All the geometric and
radiometric distortions are corrected and reduced by the application of appropriate
digital techniques which can be summarized as shown in figure (5-1).

Figure: (5-1) Satellite Image Processing.

| P a g e ( 2 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

5.2Radiometric Correction

Ideally, the radiant flux recorded by a remote sensing system in various

bands is an accurate representation of the radiant flux actually leaving the feature
of interest (e.g., soil, vegetation, water, or urban land cover) on earth‘s surface.
Unfortunately, noise (error) can enter the data collection system at several points.
For example, radiometric error in remotely sensed data may be introduced by the
sensor system itself when the individual detectors do not function properly or are
improperly calibrated. Second, the intervening atmosphere between the terrain of
interest and the remote sensing system can contribute so much noise that the
energy recorded by the sensor does not resemble that which was reflected or
emitted by the terrain.
5.3Geometric Correction

All remote sensing imagery are inherently subject to geometric distortions.

These distortions may be due to several factors, including: the perspective of the
sensor optics; the motion of the scanning system; the motion of the platform; the
platform altitude, attitude, and velocity; the terrain relief; and, the curvature and
rotation of the Earth.
Geometric corrections are intended to compensate for these distortions so that
the geometric representation of the imagery will be as close as possible to the real
world. The geometric distortions may be divided into two classes:
1. Systematic distortion: those that can be corrected using data from platform
ephemeris and knowledge of internal sensor distortion.
2. Nonsystematic distortion: those that cannot be corrected with acceptable
accuracy without a sufficient number of ground control points, which are
used to establish a relationship between the ground and image.
5.3.1 Systematic distortions

The systematic distortions of the image can be classified as:

1- Scan Skew: Caused by the forward motion if 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

| P a g e ( 3 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

2- Mirror-Scan Velocity Variance: The mirror scanning rate is usually

not constant across a given scan, producing along-scan geometric
distortion.
3- Panoramic Distortion: The ground area imaged is proportional to the
tangent of the scan angle rather than to the angle itself. Because data
are sampled at regular intervals, this produces along-scan distortion.
(Panoramic distortion is a problem common to all remote imaging systems. It is a
consequence of scanning a surface from a fixed point or path. Figure (5-4) below
illustrates the problem. In all such scanners, the instantaneous field of view (IFOV) - the
angle ϕ- is kept constant. Consequently, the width of a pixel varies depending on the
look angle θ. We can see that the width is smallest (w) at the nadir point and that it will
increase for larger values of θ).

Figure (5-4) effect of look angle on pixel width

4- Platform Velocity: If the speed of the platform changes, the ground

track covered by successive mirror scans changes, producing along-
track scale distortion
5- Earth Rotation: Earth rotates as the sensor scans the terrain. This
results in a shift of the ground swath being scanned, causing along-
scan distortion.
6- Perspective: For some applications it is desirable to have images
represent the projection of points on Earth on a plane tangent to Earth
with all projection lines normal to the plan. This introduces along-
scan distortion.

| P a g e ( 4 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

5.3.2 Nonsystematic distortions

The nonsystematic distortions of the image can be classified as:

1- Altitude. If the sensor platform departs from its normal altitude or the
terrain increases in elevation, this produces changes in scale.

2- Attitude. One sensor system axis is usually maintained normal to earth‘s

surface and the other parallel to the spacecraft’s direction of travel. If the
sensor departs from this attitude, geometric distortion results.
5.4 Image Rectification

Remotely sensed image data are gathered by a satellite or aircraft represent

the irregular surface of the Earth. Even images of seemingly flat areas are distorted
by both the curvature of the Earth and the sensor being used. This section covers
the processes of geometrically correcting an image so that it can be represented on
a planar surface, conform to other images, and have the integrity of a map.
Rectification is the process of transforming the data from one grid system into
another grid system using a geometric transformation. From above, the
rectification or registration involves the following general steps, regardless of the
application:
1. Locating Ground Control Points (GCPs).
2. Computing and testing a transformation between the original and corrected
coordinate system.
3. Creating an output image with the new coordinate. The pixels must be
resampled to conform to the new grid.

The two common geometric correction procedures often used by earth

scientists: image to-map rectification and image-to-image registration.
a. Image- to-map rectification. It is the process by which the geometry of an
image area is made planimetric by selecting ground control points from
corrected topographic map, which represent the same scene of the study area
shown in the original image.

b. Image-to-image registration. It is used when images of one area that are

collected from different sources must be used together. For comparing

| P a g e ( 5 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

separate images pixel by pixel, the pixel grids of each image must conform
to the other images in the database. Registration is the process of making an
image conforms to another image by selecting ground control points from
old corrected imagery for the same scene of the study area.
5.5 Ground Control Points (GCPs)

A ground control point (GCP) is a location on the surface of the Earth (e.g.,
a road intersection) that can be identified on the imagery and located accurately
on a map. Each GCP has two coordinates system, first the coordinates of input
image denoted by (X, Y) given in pixel unit, second the coordinates of map
denoted by (E, N) given in meter unit. The (GCPs) data can be gathered from a
wide variety of sources, including map, aerial photograph, field survey, and
Global Positioning System (GPS).
5.5.1 Selection and minimum number of GCPs

The quality of ground control points (GCPs) directly affects the accuracy of
mathematical model, and that, in turn, determines the outcome of the project.
The typical GCPs are highway intersection, airport runways, and towers that
can be clearly identified, accurately located on the image or the map. However
care must also be given to the locations of the points. A general rule is that there
should be a distribution of control points around the edges of the image to be
corrected with a scattering of points over the body of the image.
The minimum number of GCP’s required to perform a transformation of
order (t) for polynomial model equals:
�(𝑡𝑡 + 1)(𝑡𝑡 + 2)�
minimum number of GCP’s =
2
For example table (5-1) shows the minimum number of ground control
points required to perform a transformation for (1st) through (4th) order
transformation after applying above equation.

| P a g e ( 6 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

Table (5-1) Minimum number of GCPs per order of transformation.

Min. NO. of
t Order
GCP’s
3 1
6 2
10 3
15 4

However, because locating GCPs is often imprecise, it best to have many

extra points, may be twice the number of GCP required to evaluate (t) order
transform.
5.6Geometric Correction Models (transformations)

A simple geometric model usually requires mathematical functions to relate

the image space and object space. The general form of the functions can be written
as:
𝑥𝑥 = 𝑓𝑓1(𝑋𝑋, 𝑌𝑌)
𝑦𝑦 = 𝑓𝑓2(𝑋𝑋, 𝑌𝑌)
Where (x, y) are the image coordinates, (X, Y) are the object coordinates and (f1,
f2) are the mathematical functions which perform the relation between the image
and object space.
The mathematical function parameters are solved with the help of the GCPs
collected throughout the image. Once the mathematical function parameters are
determined, the correct positions of each pixel in the image can be estimated by
these functions.
5.6.1 Polynomial Transformations

Polynomial transformation usually can be used in the transformation

between image coordinates and object coordinates. The needed transformation can
be expressed in different orders of the polynomials based on the distortion of the
image, the number of GCPs and terrain type. The order of the transformation is
simply the highest exponent used in the polynomial. The first order transform is
the simple technique in the geometric correction process, it needs the minimum
| P a g e ( 7 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

No. of GCPs over other techniques. The elementary transformation such as the
rotation, shift and scale are accomplished by the first order polynomial model. The
following equations are used to express the general form of the first order
polynomial model:
𝑥𝑥 = ao + a1 X + a2 Y
𝑦𝑦 = 𝑏𝑏𝑜𝑜 + 𝑏𝑏1 𝑋𝑋 + 𝑏𝑏2 𝑌𝑌
Where (x, y) are coordinates of the GCP in the original input image while (X, Y)
represent corresponding coordinates of the GCP on the ground or map and (a, b)
are polynomial coefficients.
5.7Root Mean Square Error (RMSE)

A way to measure the accuracy of a geometric rectification algorithm

(actually, its coefficients) is to compute the Root Mean Squared Error (RMS error )
R R

for each ground control point. RMS error is the distance between the input (source)
location of a GCP and the retransformed location for the same GCP. In other
words, it is the difference between the desired output coordinate for a GCP and the
actual output coordinate for the same point, when the point is transformed with the
geometric transformation. RMS error is calculated with a distance equation:
𝑅𝑅𝑅𝑅𝑅𝑅𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 = �(𝑥𝑥𝑟𝑟 − 𝑥𝑥𝑖𝑖 )2 + (𝑦𝑦𝑟𝑟 − 𝑦𝑦𝑖𝑖 )2

Where:
xi and yi are the input source coordinates.
xr and yr are the retransformed coordinates.
RMS error is expressed as a distance in the source coordinate system.
Residuals are the distances between the source and retransformed coordinates in
one direction. The X residual is the distance between the source X coordinate and
the retransformed X coordinate. In the same manner, the Y residual is the distance
between the source Y coordinate and the retransformed Y coordinates as shown in
figure (5-5). The RMS error of each point is calculated with a distance formula:

𝑅𝑅𝑖𝑖 = �𝑋𝑋𝑋𝑋𝑖𝑖 2 + 𝑌𝑌𝑌𝑌𝑖𝑖 2

Where:
Ri = the RMS error for GCPi
XR i = the X residual for GCPi
R R

YR i = the Y residual for GCPi

R R

| P a g e ( 8 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

Fig. (5-5) Residuals and RMS Error per Point.

From the residuals, the following calculations are made to determine the
total RMS error, both of the (X RMS) error, and the (Y RMS) error need to be
calculated.

𝑛𝑛
1
𝑅𝑅𝑥𝑥 = � � 𝑋𝑋𝑋𝑋𝑖𝑖 2
𝑛𝑛
𝑖𝑖=1

𝑛𝑛
1
𝑅𝑅𝑦𝑦 = � � 𝑌𝑌𝑌𝑌𝑖𝑖 2
𝑛𝑛
𝑖𝑖=1

𝑇𝑇 = �𝑅𝑅𝑥𝑥 2 + 𝑅𝑅𝑦𝑦 2
Where:
Rx = X RMS error.
Ry = Y RMS error.
T = total RMS error.
n = the number of GCPs.
i = GCP number.
XR i = the X residual for GCPi.
R R

YR i = the Y residual for GCPi.

R R

In most cases, it is advantageous to tolerate a certain amount of error rather

than take a more complex transformation. The amount of RMS error that is
tolerated can be thought of as a window around each source coordinate, inside

| P a g e ( 9 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

which a retransformed coordinate is considered to be correct (that is, close enough

to use) as shown in figure (5-6) . For example, if the RMS error tolerance is two,
then the retransformed pixel can be two pixels away from the source pixel and still
be considered accurate.

Fig. (5-6) RMS Error Tolerance.

It is important to remember that RMS error is reported in pixels. Therefore,

if you are rectifying Landsat TM data and want the rectification to be accurate to
within 30 meters, the RMS error should not exceed 1.00.
5.8 Resampling methods

In order to actually geometrically correct the original distorted image, a

procedure called resampling is used to determine the digital values to place in the
new pixel locations of the corrected output image. 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: nearest
neighbor, bilinear interpolation, and cubic convolution.

5.8.1 Nearest Neighbor

This approach 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 the
simplest method 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, as shown in figure (5-7).

| P a g e ( 10 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

Fig. (5-7) Nearest Neighbor Resampling.

5.8.2 Bilinear Interpolation

This method takes a weighted average of four pixels in the

original image nearest to the new pixel location. The averaging process alters the
original pixel values and creates entirely new digital values in the output image.
This may be undesirable if further processing and analysis, such as classification
based on spectral response, is to be done. If this is the case, resampling may best be
done after the classification process. The weighted average of the new brightness
value (𝐵𝐵𝐵𝐵𝑤𝑤𝑤𝑤 ) is computed according to the equation:
𝑍𝑍
∑4𝑘𝑘=1 𝑘𝑘2
𝐷𝐷𝑘𝑘
𝐵𝐵𝐵𝐵𝑤𝑤𝑤𝑤 =
1
∑4𝑘𝑘=1 2
𝐷𝐷𝑘𝑘
Where (𝑍𝑍𝑘𝑘 ) are the surrounding four data point values, and �𝐷𝐷𝑘𝑘 2 � are the distances
squared from the point to these data points, as shown in figure (5-8).

| P a g e ( 11 -12)
University of Technology
Building & Construction Department / Remote Sensing & GIS lecture

Fig. (5-8) Bilinear Interpolation Resampling.

5.8.3 Cubic Convolution

This method goes even further to calculate a distance weighted average of a

block of sixteen pixels from the original image which surround the new output
pixel location, as shown in figure (5-9). As with bilinear interpolation, this method
results in completely new pixel values. However, these two methods both produce
images which have a much sharper appearance and avoid the blocky appearance of
the nearest neighbor method.

Fig. (5-9) Cubic Convolution Resampling.

| P a g e ( 12 -12)