Auto-tuning of PID Controller

Submitted in partial fulfillment of the requirements

of the degree of

Bachelors of Engineering
by

Mr. Rahul S. Roy

Roll No. 12IN2015

Supervisor

Prof. Sharad P. Jadhav

Department of Instrumentation Engineering

Dr. D. Y. Patil Groups

Ramrao Adik Institute Of Technology

Dr. D. Y. Patil Vidyanagar, Sector 7, Nerul, Navi Mumbai 400706.
(Affiliated to University of Mumbai)

2014

Declaration
I declare that this written submission represents my ideas in my own words and where
others ideas or words have been included, I have adequately cited and referenced the original
sources. I also declare that I have adhered to all principles of academic honesty and integrity
and have not misrepresented or fabricated or falsified any idea/data/fact/source in my submission. I understand that any violation of the above will be cause for disciplinary action by the
Institute and can also evoke penal action from the sources which have thus not been properly
cited or from whom proper permission has not been taken when needed.

..............................
(Signature)

..............................
(Mr. Rahul S. Roy and Roll No. 12IN2015)

Date :

Project Report Approval for B. E.

This project report entitled Auto-tuning of PID Controller by Mr. Rahul S. Roy is
approved for the degree of Bachelors Degree in Instrumentation Engineering, University of
Mumbai.

Examiners :
1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Supervisors :
1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chairman
..............................

Date :
Place :

Acknowledgments
With great pleasure, I avail this opportunity to express my profound gratitude and deep
regards to my project co-ordinator Mr.Shirish S. Kulkarni for their spirited guidance, monitoring and constant encouragement throughout the completion of this seminar report. I have
deep sense of admiration for their innate goodness and inexhaustible enthusiasm, it helped me
to work in right direction to attain desired objective.
I am also thankful to Mr.Sharad P. Jadhav, Head of Department of Instrumentation Engineering, RAIT, Nerul for his generous support, devoting their valuable time and helped me
in all possible ways towards successful completion of this work. I thank all those who have
contributed directly or indirectly to this work.
I take this privilege to express my sincere thanks to Dr.Ramesh Vasappanavara, Principal,
RAIT for their support, encouragement and providing the much necessary facilities. I extend
thanks to my friends who have supported in every stage of these report. I cannot end without
thanking my lovely family for their encouragement.

Date

Signature

Abstract
Image registration, or geometric alignment of two-dimensional or three-dimensional
(3-D) image data, is becoming increasingly important in diagnosis, treatment planning, functional studies, computer-guided therapies and in biomedical research. Image registration is an
fundamental and tough task in image processing technique. Especially, CT and MR imaging
for diagnosis and surgical planning indicates that physicians and surgeons gain important information from these modalities, this dissertation work focuses on the implementation of image
registration technique.
In medical image processing field, some techniques are proposed to find a geometrical
transformation that relates the points of an image to their corresponding points of another
image. This report deals with the use of phase correlation for estimation of mutual translation
between images. Log-Polar Transform (LPT) method will be also proposed, thus improving the
accuracy and precision of feature-based registration which will include mutual rotation, scale
and translation. A key feature of LPT methods is the speed offered by the use of FFT routines.
The proposed method will estimates large motions accurately and robustly without the need of
excessive zero padding and over-sampling, thus without sacrificing part of the computational
efficiency for the frequency domain formulation.

Contents
1 Introduction

1.1

Image Registration Concept . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2

Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.3

Organization Of Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 Literature Review

3 Image Registration using Phase Correlation

3.1

Phase Correlation Technique . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.2

Principle of PCM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.3

Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.4

Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 Performance Results

12

4.1

Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

4.2

Steps for Image Registration by PCM . . . . . . . . . . . . . . . . . . . . . . 13

5 Conclusion

14

Bibliography

15

ii

List of Figures
1.1

Image Registration Concept [1] . . . . . . . . . . . . . . . . . . . . . . . . . .

iii

List of Tables
3.1

Calculation Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

iv

Chapter 1
Introduction
Medical images are increasingly widely used in healthcare and bio-medical research from
wide range of imaging modalities which are now available, such as CT, MRI, PET, SPECT, and
so on. In some clinical scenarios, the information from several different imaging modalities
should be integrated to deduce useful clinical conclusions. Image registration aligns the images
and so establishes correspondence between different features contained on different imaging
modalities, allows monitoring of subtle changes in size or intensity over time or across a population and establishes correspondence between images and physical space in image guided
interventions [2]. Of the multitude of Image registration similarity measures that have been
proposed over the years, mutual information is currently one of the most intensively researched
measures. This attention is a logical consequence of both the favourable characteristics of
the measure and the good registration results reported. Mutual information is an automatic,
intensity-based metric, which does not require the definition of landmarks or features such as
surfaces and can be applied in retrospect. Furthermore, it is one of the few intensity-based
measures well suited to registration of multimodal images.

1.1 Image Registration Concept

Image registration is the process of determining the spatial transform that maps points
from one image (defined as the moving image) to homologous points on an object in the ref-

Auto-tuning of PID Controller

erence image (called as the fixed image). The similarity of the two images will be calculated
and investigated after each transform until they are matched [1] [4]. It is a process of aligning
two images into a common co-ordinate system thus aligning them in order to monitor subtle
changes between the two as shown in figure 1.1. Registration algorithms compute transformations to set correspondence between the two images.
In image processing, often (for example, when comparing or combining the information
content of images), interest is in the relationship between two or more images. The analysis of this relationship usually becomes tractable once a correspondence is set up between the
images. Image registration is the task of setting up this correspondence. Image registration
shows up in a rich range of application domains, such as medical image analysis (e.g. diagnosis), neuroscience (e.g. brain mapping), computer vision (e.g. stereo image matching for
shape recovery), astrophysics (e.g. the alignment of images from different frequencies), military applications (e.g. target recognition), etc. The definition of correspondence varies across
disciplines and even across individual applications.

1.2 Proposed Method

Medical Image registration requires a high degree of precision. In order to improve the
accuracy of feature based registration, this project proposes a Log- Polar Transform (LPT)
based image registration. Instead of using the whole image in the conventional method, feature
points will be used in this project, which reduces the computational time. For rotation invariant, it is not important how the image patch is rotated. The key is focusing on the feature points.
The proposed process in this project contains four steps. First, feature points are selected in
both the reference image and the sensed image with corner detector. Secondly, image patches
will be created using feature point positions as centers. Each point is a center point of LPT, so
circular image patches are cropped choosing feature point as center. The radius of the circle
can be changed, then the circular images are transformed to log-polar coordinates. Next, the
LPT images are dealt with using phase correlation.

Ramrao Adik Institute of Technology

Auto-tuning of PID Controller

Figure 1.1: Image Registration Concept [1]

1.3 Organization Of Report

The report is organized as follows. Chapter 1 consists of introduction, image registration
concept and process. Chapter 2 gives brief information of literature review. Chapter 3 constitutes of phase correlation technique, strategy of implementation. Performance result are
included in Chapter 4. Chapter 5 concludes the report. Finally, bibliography section contains
an overview of the previously published and referred conference and journal papers as well as
books.

Ramrao Adik Institute of Technology

Chapter 2
Literature Review
Batler et al. registered 2-D projection images using a curve matching method [2]. In these
method samples are generated to sequences of corresponding points to be registered. Corresponding open curves are manually identified, then are matched by searching for the optimal
fit of the local curvatures in the two curves which leads to registration of two-dimensional
projection radiographies. Nicholas Ayache and Andre P. Gueziec have presented a method of
matching of 3D objects extracted from medical images. Crest lines computed on the object
surfaces correspond to meaningful anatomical features, and that they are stable with respect
to rigid transformations. The extraction of the crest lines is done by computing up to third
order derivatives of the image intensity function with appropriate 3D filtering of the volumetric
images, and by the marching lines algorithm. The recovered lines are then approximated by
spline curves, to compute at each point a number of differential invariants. Matching is finally
performed by a new geometric hashing method. The whole chain is now completely automatic,
and provides extremely robust and accurate results, even in the presence of severe occlusions.
Peng Wen [3] , has suggested a medical image registration method using points, contour and
curves is proposed, which has the accuracy of registration based-on points and the robust of
registration based-on lines (including contour and curves). The operator can exactly extract the
features from the images with semi-automatic extraction method.
Audette and Ferrie [4] have reviewed the surface based registration methods in medical

Auto-tuning of PID Controller

imaging. Boundaries or surfaces, in medical images are many times more distinct than landmarks and hence can be used for segmentation by locating high contrast surfaces. Surface
matching algorithms are normally used for rigid body registration. The representation of surface methods is a surface-based approach for registering multimodality brain image. They fit
a set of points extracted from contours in one image to a surface model and extracted from
contours in the other image. The image that covers the larger volume of the patient, or the
image that has a higher resolution if volume coverage is comparable, is used to generate the
surface model. Another version of surface matching is to provide user with an interactive
transformation package that allows the user to translate and rotate one image with respect to
the other. Jack and Roux have proposed to formulate the surface registration problem as a
high dimensional optimization problem, which can be solved by a genetic algorithm. Pelizarri
had proposed a surface fitting technique for registration of images of the head which became
known as Head and Hat Algorithm. The registration algorithms have been developed in recent
years are Iterative Closest Point algorithm and Correspondence matching.
If one image is acquired from a single patient, and the other image is taken from an image
information database, we call it atlas registration. In the literature, registration of a patient
image to an image of a normal subject is termed atlas registration. This also may be useful
from the point of view of research studies viz. study of images obtained from separate individuals. Wan Rui and Li Minglu [5] have stated that standardized brain coordinate system
has been used in automatically segmenting individual cerebral structures in brain image volumes and stereotactic neurosurgical planning. These techniques have been proved to be useful
in medical diagnosis. K.K. Bhatia, J.V. Hajnal, B.K. Puri, A.D. Edwards and D. Rueckert
have described group wise registration algorithm that simultaneously registers all images to
a common reference space[6]. Gradient projection method has been applied constrained optimization is applied to maximize the similarity between the images. Daniel J. Blezek and James
V. Miller have explored question, whether a single atlas is appropriate for a given sample or
whether there is an image based evidence from which we can infer multiple atlases, each constructed from a subset of the data, referring this process as atlas stratification.
Moments are used in classical mechanics to characterize rigid bodies by the spatial distriRamrao Adik Institute of Technology

Auto-tuning of PID Controller

bution of their mass. The principal axes are those orthogonal axes about which the moments
of inertia are minimized. If two objects are identical except for a translation and a rotation,
then they can be registered exactly by bringing their principal axes into coincidence. If two
objects are shaped similarly, they may be registered approximately by this process. J.Flusser
and T. Suk derived affine transform invariants and subsequently used it for registration of SPOT
and Landsat images [7]. Moment based methods also appear as hybridly classified registration
methods that use segmented or binarized image data for input. In many applications, presegmentation is mandatory in order for moment based methods to produce acceptable results.
Mutual information based registration begins with the estimation of the joint probability of
the intensities of corresponding voxels in the two images. Registration of multi-modal images
is the difficult task, but often necessary to solve, especially in medical imaging. The comparison of anatomical and functional images of the patients body can lead to a diagnosis, which
would be impossible to gain otherwise. Remote sensing often makes use of the exploitation
of more sensor types. P. Viola and W.M. Wells have maximized mutual information using
gradient descent optimization method [8]. The use of information-theoretic measures such
as mutual information has obviously benefited voxel-based registration. The present papers
have demonstrated that mutual information can be used to parameterize and solve the correspondence problem in feature-based registration. They have appeared recently and represent
the leading technique in multimodal registration. Researchers have used various methods like
Parzen window, spline pyramids to achieve mutual information based image registration. Frederik Maes, Dirk Vandermeulen and Paul Suetens have investigated the performance of various
optimization methods and multiresolution strategies for maximization of mutual information.
Josien P. W. Pluim, J. B. Antoine Maintz and Max A. Viergever have given a review of mutual
information based image registration methods.

Ramrao Adik Institute of Technology

Auto-tuning of PID Controller

Correlation methods are useful for registration of monomodal images, for comparison of
several images of the same object, e.g. to analyze development of the disease. The early survey
paper by K. Ghaffary, A.A. Sawchuk covers mainly the methods based on image correlation.
The most exhaustive review of the general-purpose image registration methods has been given
by Gotsfeld. Literature on cross-correlation of original images or extracted feature images is
found in literature on fourier domain based cross-correlation and phase-only correlation. Min
Xu and Varshney P.K. have specified the application of a subspace-based frequency estimation
approach for the Fourier based image registration problem [7]. They have used multiple signal
classifier algorithm for more robust and accurate results resulting into moderate increase in
computational complexity.
Correlation technique was used to refine the user-defined landmark positions to register
tomographic brain and abdominal images. A. Collignon has suggested the use of the Entropy
Correlation Coefficient (ECC), for image registration using normalized mutual information.
S. Sanjay-Gopal, H. P. Chan, T. Wilson, M. Helvie, N.Petrick and B. Sahiner have compared
mutual information and the correlation coefficient for registration of intra subject mammograms [9]. Samritjiarapon O. and Chitsobhuk O. have used fourier-based technique along with
best-first search algorithm to find the translation between two input images. The Fourier-based
technique is used to estimate the candidate translations to decrease searching space while bestfirst search algorithm is used to further search for the correct translation. This technique can
estimate large translations, scalings, and rotations in images by an extension of phase correlation technique. They focus on increased accuracy of this technique to detect large translations
compared to the other techniques in frequency domain.

Ramrao Adik Institute of Technology

Chapter 3
Image Registration using Phase
Correlation
Phase correlation measures the local movement between two images from their phases
with sub pixel accuracy. These operation results not only higher registration accuracy but also
lower computational complexity.

3.1 Phase Correlation Technique

In image processing, phase correlation is a method of Image registration, and uses a fast
frequency-domain approach to estimate the relative translative offset between two similar images. The ability to detect and estimate lateral shifts between similar images is an integral part
of many image processing applications. Many methods have been developed to estimate the
translational bulk displacement between similar images. The phase correlation method (PCM)
[8] is a popular choice due to its robust performance and computational simplicity. This method
is a natural fit with MRI because the acquired image data is typically sampled in the spatial
fourier domain. Thus, image registration using phase correlation can be applied directly to the
raw data before the spatial images are reconstructed.
The PCM is based on the well-known fourier shift property. Specifically, a shift in the

Auto-tuning of PID Controller

coordinate frame of two functions results in a linear phase difference in the fourier transform
of the two functions [7]. Given a pair of two-dimensional (2-D) functions, A and B, that are
related by a simple translational shift, the elements of the fourier transform of B, are related to
A by;

B(k,l) = A(k,l)exp-j(ka+lb)

(1)

where (k,l) are the fourier domain coordinates, and a and b are the magnitude of the horizontal and vertical shifts that occur between A and B.
To identify a and b using the PCM approach, one computes a normalized cross power
spectrum between A and B to identify the phase correlation matrix

B(k,l)A(k,l)
Q(k,l)= |B(k,l)A(k,l)
| =exp-j(ka+lb)

(2)

Table 3.1: Calculation Table

Date

Price

Size

Yesterday

big

Today

small

3.2 Principle of PCM

Suppose that there is a shift (d1 , d2 ) between two images sk and sk+1 :
sk (n1 , n2 ) = sk+1 (n1 + d1 , n2 + d2 )

(1)

The shift in the spatial-domain is thus reflected as a phase change in the frequency domain.
Thus by obtaining complex-valued cross-power spectrum expression
Ck,k+1 (f1 , f2 ) = Sk+1 (f1 , f2 )Sk (f1 , f2 )
Ramrao Adik Institute of Technology

(2)
9

Auto-tuning of PID Controller

where denotes the complex conjugate. To reduce the influence of luminance variation,
the right side of (2) is normalizes as follows:
((Ck,k+1 (f1 , f2 )) =

Sk+1 (f1 ,f2 )Sk (f1 ,f2 )

|Sk+1 (f1 ,f2 )Sk (f1 ,f2 )|

(3)

From (1) and (3), following equation can be obtained,

ck,k+1 (n1 , n2 ) = (n1 d1 , n2 d2 )

(4)

where denotes the pulse function.

The final equation indicates that if we can find the location of the pulse in the crosscorrelation map of two images, then the spatial displacement is be easily obtained between
them. It is shown in cross correlation map in fig 3.1.

3.3 Strategy
The illuminating feature is that the problem of finding the exact lateral shift between two
images is recast as finding the rank-one approximation of the normalized phase correlation
matrix, Q. A close inspection reveals that the noise-free model for Q is in fact a rank-one
matrix. Each element in Q can be separated as;
Q(k,l)= exp(-jka)exp(-jlb)

(3)

This allows the definition of two vectors

qa (k)=exp(-jka) and qb (l)=exp(+jlb)

(4)

and the phase correlation matrix can be rewritten as Q = qa qbH . A straightforward approach
to finding the dominant rank-one subspace of Q is to use the singular value decomposition
(SVD). The linear phase coefficients can then be identified independently in the left and right
dominant singular vectors. From these, estimates of the vertical and horizontal shift can be
derived, even for noninteger translational motion over a large range. To identify the linear
phase in each of the right and left dominant singular vectors of Q.
Ramrao Adik Institute of Technology

10

Auto-tuning of PID Controller

3.4 Applications
The advantage of computing frequency domain phase correlation is speed. A spatial domain phase correlation requires a significantly larger number of multiplications due to the
added step of the convolution operation. The convolution, as translated in the frequency domain is a simple point by point multiplication leading to a large reduction in the required number of multiplication[12]. The addition of real-time correlation in medical imaging systems to
allow the examination of 3D and 4D images presents significant benefits to all involved, and
thus can help guide medical professionals in the use and adoption of these new technologies.
Image registration using phase correlation finds its applications [7] in medicine combining
data from different modalities e.g. computed tomography (CT) and magnetic resonance imaging (MRI), to obtain more complete information about the patient, monitoring tumor growth,
treatment verification, comparison of the patients data with anatomical atlases[6], in cartography (map updating), remote sensing (multi-spectral classification), integrating information
into geographic information systems (GIS), and in computer vision. It demonstrates superior
performance in identification of cancerous lesions with the use of registration.

Ramrao Adik Institute of Technology

11

Chapter 4
Performance Results
The Fourier transform has several properties that can be exploited for image registration. Translation, rotation, reflection and scale all have their counterpart in Fourier domain[3].
Moreover, using the frequency domain, it is possible to achieve excellent robustness against
correlated and frequency dependent noise. An elegant method used to register two images
which are shifted relative to one another is to use phase correlation [9]. Phase correlation relies
on the translation property of the Fourier transform better known as the shift theorem. The
algorithm in figure 4.1 has been implemented using MATLAB software. Performance analysis
has been done at each stage and the results are evaluated and explained below. We tested the
algorithm on different images.

4.1 Algorithm
Given two images f1 and f2 which differ by translation (dx, dy), i.e.;
f2 (x, y) = f1 (x dx, y dy)

(1)

Their corresponding Fourier transform F1 and F2 will be related by;

F2 (x , y ) = ex dx +y dy F1 (x , y )

(2)

The principle used of correlation is for matching. In matching, f1 (x, y) is an image containing objects or regions of interested and let f2 (x, y) be that object or region (template). Then, if
12

Auto-tuning of PID Controller

there is a match, the correlation of the two functions will be maximum at the location where h
finds a correspondence f.
The correlation of two function f1 (x, y) and f2 (x, y) is defined as;
f1 (x, y) f2 (x, y) =

1
MN

PM 1 PN 1
m=0

m=0

f1 (m, n) f2 (x + m, y + n)

(3)

4.2 Steps for Image Registration by PCM

First two input images I1 (x, y) and I2 (x, y) are loaded.

2D-FFT of both the images is calculated. This step will give two 2-D arrays of FFT
co-efficient I1 and I2 . Then the magnitude of FFT co-efficient is convolved.

IFFT is calculated to find out the matching point (Tpeak) having maximum magnitude.

The co-ordinates of the above obtained peak (Tpeak) will give the translation between
the images.

Thus by removing the translation from the image, will lead to registered image.

Ramrao Adik Institute of Technology

13

Chapter 5
Conclusion
In this report, algorithms to spatially register digital images is being incorporated. The
theoretical motivation for the employment of Image registration using phase correlation is provided. The method is derived by expressing a robust matching surface in terms of correlations.
The speed of the method comes from using fast fourier transforms to compute the correlations.
In this approach the transform coefficients are normalized to unit magnitude prior to computing
correlation in the frequency domain. Thus, the correlation is based only on phase information
and is insensitive to changes in image intensity. Although it has the drawback that all transform
components are weighted equally. In principle one should select the spectral pre-filtering so
as to maximize the expected correlation signal-to-noise ratio given the expected second order
moments of the signal and signal noise. The rotation and scale invariant properties of the logpolar transform, along with fast fourier transform and phase correlation allow us to develop
a robust algorithm that works faithfully under geometric distortions like rotation, scale and
translation. Hence this approach is highly effective in registering medical images.

14

Bibliography
[1] Chia-Hsiang Menq Peng Cheng. Real-time continuous image registration enabling ultraprecise 2-d motion tracking. IEEE TRANSACTIONS ON IMAGE PROCESSING,
5:20812090, 2013.
[2] Chen GT Balter JM, Pelizzari CA. Correlation of projection radiographs in radiotherapy
using open curve segments and points. Med Physics, 19:329334, 2010.
[3] Peng Wen. Medical image registration based-on points, contour and curves. International
Conference on Biomedical Engineering and Informatics, pages 132136, 2008.
[4] T.M. Peters M.A. Audette, F.P. Ferrie. An algorithmic overview of surface registration
techniques for medical imaging. Medical image Analysis, 4:201217, 2000.
[5] Prof.Li Minglu Wan Rui. An overview of medical image registration. Proceedings
of the Fifth International Conference on Computational Intelligence and Multimedia
Applications (ICCIMA03), pages 385390, 2003.
[6] B.K. Puri A.D. Edwards K.K. Bhatia, J.V. Hajnal and D. Rueckert. Consistent group
wise non-rigid registration for atlas construction. IEEE International Symposium on
Biomedical Imaging, 1:1518, 2004.
[7] Varshney P.K. Min Xu.

A subspace method for fourier-based image registration.

Geoscience and Remote Sensing Letters, IEEE 33, 6:491494, 2009.

[8] Shigeo Wada Zhuo Liu. Rotation and scale invariant texture characterization method.
IEEE proceedings of ISSPIT, pages 103106, 2004.
15

Auto-tuning of PID Controller

[9] M.Israeli Y.Keller, A.Averbuch. Pseudo polar-based estimation of large translations, rotations and scalings in images. IEEE trans. on Image processing, 26:1222, 2005.
[10] Ali Uneri Sebastian Schafer Sajendra Nithiananthan Douglas D. Evaluation of a system for high-accuracy 3d image-based registration of endoscopic video to c-arm conebeam ct for image-guided skull base surgery. IEEE TRANSACTIONS ON MEDICAL
IMAGING, 32:12151226, 2013.
[11] Dirk Vandermeulen Pieter Slagmolen, Frederik Maes. Nonrigid image registration with
crystal dislocation energy. 22:229243, 2013.
[12] Rui Liao Li Zhang, Ying Sun. A review of recent advances in registration techniques
applied to minimally invasive therapy. IEEE TRANSACTIONS ON MULTIMEDIA,
15:9831000, 2013.
[13] XuZhi Li Kai Zhang and JiuXing Zhang. A robust point-matching algorithm for remote
sensing image registration. IEEE GEOSCIENCE AND REMOTE SENSING LETTERS,
11:469473, 2014.
[14] Dr. Pradeep M. Patil Medha V. Wyawahare and Hemant K. Abhyankar. Image registration
techniques: An overview. International Journal of Signal Processing, Image Processing
and Pattern Recognition, 2:1128, 2009.
[15] Hongshi Yan and Jian Guo Liu.

Robust phase correlation based feature matching

for image co-registration and dem generation.

The International Archives of the

Photogrammetry, Remote Sensing and Spatial Information Sciences., XXXVII:Part B7,

2008.
[16] J. G. Proakis and D. Manolakis. Digital Signal Processing : Principles,Algorithms, and
Applications. New York: Prentice-Hall, 3rd edition, 1996.
[17] Gonzalez and Woods. Digital Image Processing. Prentice Hall, 3rd edition, 2008.
[18] Yushuang Tian and Kim-Hui Yap. Joint image registration and super-resolution from
low-resolution images with zooming motion. IEEE TRANSACTIONS ON CIRCUITS
AND SYSTEMS FOR VIDEO TECHNOLOGY, 23:12241234, 2013.
Ramrao Adik Institute of Technology

16

Auto-tuning of PID Controller

[19] Tie-Niu Tan Jun-Zhou Huang. Phase correlation based iris image registration model.
Journal of Computer Science and Technology, pages 419425, 2005.
[20] Salman Bombaywala Jignesh N Sarvaiya, Dr. Suprava Patnaik. Image registration by
template matching using normalized cross-correlation. IEEE International Conference
on Advances in Computing, Control, and Telecommunication Technologies, pages 819
822, 2009.
[21] Olan Samritjiarapon Orachat Chitsobhuk. An fft-based technique and best-first search
for image registration. Communications and Information Technologies, ISCIT, pages
364367, 2008.
[22] Chin-Hsing Chen Wen-Chia Lee. A fast template matching method for rotation invariance using two stage process. Intelligent Information Hiding and Multimedia Signal
Processing, pages 912, 2009.
[23] Dae Hwan Hwang Yun-Koo Chung Kyu-Dae Ban, Jaeyeon Lee. Face image registration methods using normalized cross correlation. International Conference on Control,
Automation and Systems, pages 24082411, 2008.
[24] D. Zhou A B. Sun. Rotated image matching method based on cisd. LNCS 4491, Part
I:13461352, 2007.
[25] M.Israeli Y.Keller, A.Averbuch. Pseudo polar-based estimation of large translations, rotations and scalings in images. IEEE trans. on Image processing, pages 1222, 2005.

Ramrao Adik Institute of Technology

17