See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/301360392

Geometric Distortion and Correction Methods

for Finding Key Points:A Survey

Article · April 2016

CITATIONS READS

0 248

2 authors, including:

Amit Patel
Charotar University of Science and Technology
44 PUBLICATIONS 109 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Design and development of Transmission line components 42GHz/200 kw/3 sec gyrotron for startup
studies of tokamak aditya View project

Development of Non Contact Liquid Level Measurement Device based on FMCW Radar Principle View
project

All content following this page was uploaded by Amit Patel on 19 April 2016.

The user has requested enhancement of the downloaded file.

 IJSRD - International Journal for Scientific Research & Development| Vol. 4, Issue 02, 2016 | ISSN (online): 2321-0613

Geometric Distortion and Correction Methods for Finding Key Points:A

Survey
Devangi B. Thakkar1 Asst. Prof. Amit V. Patel2 Prof. Mayur Patel3
1
M.Tech. Student
1,3 2
Department of Computer Engineering Department of Electronics & Communication Engineering
1,3
U. & P. U Patel C.S.P.I.T., Charusat 2V. T. Patel Charusat University
Abstract— In remote sensing data for georeferencing refers Platform Velocity: If the speed of the platform
to the process of assigning to image data from some map changes, the ground track wrapped by successive mirror
coordinate. Geometrical distortion is an error on an image scans changes, manufacturing along-track scale distortion
between actual image acquisition (sensor) coordinate data Earth Rotation: Earth rotates as the acquisition
and ideal image coordinate data. Geometric correction scans the terrain. This results in a shift of the ground
includes correcting for geometric distortion due to some swath being scanned, causing along-scan distortion.
variation and collecting Ground control point on that image.
B. Non Systematic Error
This review paper include preprocessing of satellite image
using sift algorithm for further application such as change These errors are corrected through the use of ground control
detection, image acquisition, image registration, etc. points (GCP’s).
Key words: Geometric distortion; Scale invariant feature Altitude Variance: If the sensor acquisition
transformation; pick signal to noise ratio; Image acquisition; platform leave from its normal altitude or the terrain
ground control point increases in promotion, this produces changes in scale or
pixel size.
I. INTRODUCTION Platform Attitude: One sensor system axis is
usually maintained normal to Earth's surface and the other
Remote sensing is defined as the measurement of object collateral to the spacecraft's direction of travel.
properties on the earth surface using acquired data from
satellite images and aircraft. There are two characteristic of
II. GEOMETRIC DISTORTION
on image: geometric correction and distortion describe on
the actual image coordination. Distortion is on noise Geometric distortion is an error on an image, between the
information and reverting of the original image and shape. undeniable image coordinates and the supreme image
Images which are more realistic generally contain such coordinates which would be predicted theoretically with a
significant amount of by changes in the shape of an image supreme sensor and under supreme conditions. Satellite
that they cannot be used personally with map base products images are mostly contorted due to, Camera and sensor
like, geographic information system (GIS) [1]. The word orientation, Systematic error, Topographic relief
image registration is used in separate literatures as displacement, Earth curvature, Motion of earth. Geometric
geometric registration, and rectification of geometric distortions are classified into internal distortion resulting
distortion affine transformation. The image acquired earlier from the geometry of the sensor, and external distortions
date is known as base image and, recently obtained image resulting from the attitude of the sensor or the form of the
with error is to be corrected. Preprocessing of satellite object. Various example of internal and external geometric
images prior to image application is essential. Pre- distortion are given in Figure 1[4] and Figure 2 [4]
processing commonly encompasses a series of sequential respectively.
operations like, atmospheric correction (normalization),
image registration, geometric correction [2].Using that of
the geometric correction we can find the distortion error in
image for reconstruction satellite images. There exist basic
two types of errors in the remotely sensed imagery which
are,
A. Systematic Error
These errors are corrected through analysis of system
characteristics and ephemerides.
Scan Skew: Caused by the forward motion if some
time required during the platform for each mirror sweep.
The ground Swath is not normal to the ground track but it
slightly skewed producing cross-scan geometrical distortion.
Mirror-Scan Velocity Variance: The mirror
scanning rate is generally not stable across a given scan,
producing along-scan geometric distortion.
Panoramic Distortion: The earth area imaged is
affinity to the tangent of the angle itself rather than scan
angle. Because data are sampled at regular intervals, it
produces along-scan deformation. Fig. 1: Internal Distortion

All rights reserved by www.ijsrd.com 311

 Geometric Distortion and Correction Methods for Finding Key Points:A Survey
Fig. 2: External Distortion
III. GEOMETRIC CORRECTION
The use of standard pixel size and coordinate permit suitable
layering of images from different sensors and map in to GIS.
The final layer of that orthorectification. On that layer the
correction is pixel by pixel on the topographic image every
resulting on the top of above if the image appears the image
is on orthographic projection.
However, Geometric correction is needed to
preprocess remotely sensed data and to eliminate geometric
distortion such as Internal and External distortion, so that
individual image pixels are in their appropriate plan metric
(x, y) map locations. This allows remote sensing derived
information to be related to other thematic information in
Geographical Information System. Geometrically corrected
imagery can be used to extract exact direction, distance and
polygon area information and find the ground control points
[3] using sift algorithm.
Fig. 3: Geometrtic Correction
(IJSRD/Vol. 4/Issue 02/2016/094)
IV. GEOMETRIC RECTIFICATION
Geometric rectification [4] is the process by which the
geometry of an image is made plan metric. It necessitate
ground control point for that map coordinates are necessary
It is an accurate process, since each pixel is not allusion of
only by its image coordinate but also by its ground
coordinate system in a standard map rejection and hence, it
is known image to map rectification. Geometric rectification
has to be done when linear accurate or expanse of space
measurements are needed. Many times, accurate geometric
rectification is not involved and in that case as a substitute
applying geometric rectification of all the images, the
images is registered with each other which is known as
image registration.
In order to harmonize a remote sensing image to a
map coordinate system, two basic operations have to be
performed. Firstly is spatial interpolation and secondly is
intensity interpolation. So we can see in spatial
interpolation, a enormous of number of GCPs is defined on
the image and on the earth. A polynomial is fitted to the
control points using minimal squares criteria to model the
geometric error undeviating into the image without
explicitly identifying their source [4].
V. ORTHORECTIFICATION
From the nadir angle, the camera is not viewing at the scene
then the geometry of a expressed images. The images have
been altering by a process known as Orthorectification [5].
Ortho-rectification, converts an image obtained at an
random roll-pitch-yaw of the camera to one as if acquired
when roll, pitch, and yaw were all zero. Ortho-rectification
overturns the camera as if it is always glance down and
heading toward the north. Ortho-rectified images are easier
to register and scrutiny because these are not geometrically
distorted. Correction for image geometry by
orthorectification, however, the intensity produces by some
geometric images.
VI. RESAMPLING
There are various common methods for re-sampling namely
nearest neighbor, Bilinear interpolation, Cubic interpolation.
In all methods, the number of rows and columns of pixels in
the output are calculated from the dimensions of the output
map, which is determined by the geometric transformation
and the cell size. The output corners side (upper left and
lower right) of the output file can be described. The default
values are calculated so that the entire source file is re-
sampled to the destination file.
A. Nearest Neighbour
Nearest neighbor re-sampling uses the in this Figure.4 [6]
digital value from the pixel in the original image which is
closed to the new another pixel location in the rectified
image. One method which is easy and does not alter the
original values but may result in some pixel values being
duplicated while others are lost. This method also inclines to
result in a disjointed image aspect.
All rights reserved by www.ijsrd.com 312
 Geometric Distortion and Correction Methods for Finding Key Points:A Survey
Fig. 4: Nearest Neighbor
B. Bilinear Tranformation
In bilinear interpolation, the data file value of the corrected
pixel is based upon the distances between in the Figure.5 [6]
retransformed coordinate location (xr, yr) and the four
nearest pixels in the source image.
Fig. 5: Bilinear Transformation
C. Cubic Convolution
This process is similar as bilinear interpolation except a set
of pixel in a 4×4 array are arranged to determine the output
data file and an approximation of a cubic function, rather
than a linear function, is applied to those 16 input values. To
identify the 16 pixels in relation to the retransformed
coordinate (xr, yr), the pixel (i, j) is used, like,
𝑖 = 𝑖𝑛𝑡(𝑥𝑟 )
𝑗 = 𝑖𝑛𝑡(𝑦𝑟 )
This assumes that (xr,yr) is expressed in Figure.6 [6] data
file coordinates (pixels). The pixels around (i,j) make up a
4×4 grid of input pixels.
Fig. 6: Cubic Convolution
Using resampling method we can apply scale feature
transformation on an image for find the key point.
(IJSRD/Vol. 4/Issue 02/2016/094)
VII. METHOD FOR FINDING KEY POINT
1) Take octave function or take matrix
2) Apply Gaussian filtering and distributed filtering on
that image
3) Extract key point
4) Apply convolution and rotate the image
VIII. CONCLUSION
Every remote sensing satellite requirements of geometric
correction process.. This paper is only review the geometric
correction. Scale feature invariant transform algorithm may
use for recognizing further relationship between images.
This paper only enlist and provide overview of various
geometric correction parameters, but one can further take
following references for more data.
REFERENCES
[1] Noam Levin, ”Fundamentals of Remote Sensing”,
Elsevier Publications, 1999, pages “57–97”
[2] Lt. Dr. S. Santosh Baboo, M. Renuka Devi, “Geometric
Correction in Recent High Resolution Satellite Imagery:
A Case Study in Coimbatore, Tamilnadu”, International
Journal of Computer Applications, January-2011, 0975-
8887
[3] Geometric Correction in High Resolution Satellite
Imagery using Mathematical Methods: A Case Study in
Kiliyar Sub Basin By Capt. Dr. S. Santhosh Baboo &
Mr. S. Thirunavukkarasu Dwaraka Doss Goverdhan
Doss Vaishnav College, India
[4] A Survey on Geometric Correction of Satellite
ImageryInternational Journal of Computer Applications
(0975 – 8887)Volume 116 – No. 12, April 2015
[5] Narayan Panigrahi, B.K. Mohan and G. Athithan, “Pre-
processing Algorithm for Rectification of Geometric
International Journal of Computer Applications
(09758887) Volume 116 – No. 12, April 2015.
[6] ERDAS Field Guide™ Fifth Edition, Revised and
Expanded ORTHORECTIFICATION OF HIGH
RESOLUTION SATELLITE IMAGES P. Boccardo a ,
E. Borgogno Mondinoa,*, F. Giulio Tonolo a, A.
Lingua a Politecnico di Torino, Dipartimento di
Georisorse e Territorio, C.so Duca degli Abruzzi 24,
10129 Torino, ITALY (piero.boccardo,
enrico.borgogno, fabio.giuliotonolo,
andrea.lingua)@polito.it
[7] SELECTION OF THE MOST SUITABLE SIZES OF
GROUND CONTROL POINTS INTHE SATELLITE
IMAGES H.M. Yilmaz a, M. Yakar b, O. Mutluoglu b,
F. Yildiz b a Nigde University, Engineering and
Architecture Faculty, Geodesy and Photogrammetry
Department, Nigde, TURKEY (hmyilmaz@nigde.edu.tr
b Selcuk University, Engineering and Architecture
Faculty, Geodesy and Photogrammetry Department,
42075 Konya, TURKEY Commission IV, WG IV
[8] Effect of Ground Control points Location and
Distribution Geometric Correction Accuracy of Remote
Sensing Satellite ImagesF. ELtohamy*, E. H. Hamza**
13th International Conference on AEROSPACE
SCIENCES & AVIATION TECHNOLOGY,ASAT-
13, May 26 – 28, 2009, E-Mail: asat@mtc.edu.eg
Military Technical College, Kobry Elkobbah, Cairo,
All rights reserved by www.ijsrd.com 313
 Geometric Distortion and Correction Methods for Finding Key Points:A Survey
(IJSRD/Vol. 4/Issue 02/2016/094)

Egypt Tel : +(202) 24025292 – 24036138, Fax: +(202)

22621908
[9] Optimization of Geometric Correction of Landsat TM
Images using Digital Elevation Model Yoshikazu Iikura
Faculty of Science and Technology, Hirosaki
University 3 Bunkyo-cho, Hirosaki 036-8561 Japan.
[10] Automation of Precise Ortho-rectification of Landsat
TM Images Yoshikazu Iikura Faculty of Science and
Technology, Hirosaki University 3 Bunkyo-
cho,Hirosaki 036-8561 Japan iikura@cc.birosaki-
u.ac.jp
[11] Tracking a Person with Pre-recorded Image Database
and a Pan, Tilt, and Zoom Camera Yiming Ye IBM T.J
Watson Research Center P.O. Box 704,Yorktown
Heights, N.Y. 10598
[12] Implementing the Scale Invariant Feature
Transform(SIFT) Method YU MENG and Dr. Bernard
Tiddeman(supervisor) Department of Computer Science
University of St. Andrews yumeng@dcs.st-and.ac.uk
[13] Scale Invariant Feature Transform (SIFT): Performance
and Application Vedrana Andersen, Lars Pellarin and
Ren´ee Anderson June 4, 2006
[14] SUSAN—A New Approach to Low Level Image
Processing STEPHEN M. SMITH¤ Department of
Clinical Neurology, Oxford University, Oxford, UK
steve@robots.ox.ac.uk J. MICHAEL BRADY
Department of Engineering Science, Oxford University,
Oxford, UK jmb@robots.ox.ac.uk Received May 3,
1993; Revised February 3, 1995; Accepted October 23,
1995
[15] Validation of an automatic image ortho-rectification
processing S. BAILLARIN, J.P. GLEYZES, C.
LATRY CNES (Centre National d'Etudes Spatiales)
Toulouse, France simon.baillarin@cnes.fr C. VESCO
Spot Image TOULOUSE, France A. BOUILLON, E.
BRETON, L. CUNIN French National Geographic
Institute Space Department Ramonville- France
aurelie.bouillon@cnes.fr J.M. DELVIT Capgemini
Space Unit TOULOUSE, France
[16] Ground Truth Extraction from LiDAR Data for Image
Orthorectification Xiaoye Liua, b, Zhenyu Zhang a, b a
Centre for GIS, School of Geography and
Environmental Science Monash University Clayton,
Victoria 3800, Australia b Faculty of Engineering and
Surveying University of Southern Queensland
Toowoomba, Queensland 4350, Australia
liux@usq.edu.au.

All rights reserved by www.ijsrd.com 314

View publication stats