IRJET- A Comparative Review of Satellite Image Super Resolution Techniques
0 likes•10 views
This document presents a comparative review of satellite image super resolution techniques in the frequency domain. It discusses several common interpolation and wavelet transform based methods for enhancing the resolution of satellite images, including nearest neighbor interpolation, bilinear interpolation, bicubic interpolation, 2D wavelet transformation with and without wavelet zero padding and cycle spinning, complex wavelet transform, and discrete wavelet transform. The review finds that while interpolation alone is insufficient, combining interpolation like bicubic interpolation with wavelet transforms provides better performance for enhancing satellite image resolution.
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1465
A Comparative Review of Satellite Image Super resolution techniques
Kangan Sardar1, Inderpreet Kaur2
1 M.Tech student, Computer science and Engineering, Bahra Group of Institutions, Patiala, India
2 Assistant Professor, Computer science and Engineering, Bahra Group of Institutions, Patiala, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract –Resolution of satellite imagesisaveryimportant
factor and needs to be preserved for various applications. A
number of techniques have been proposed by various
researchers in this context. Image resolution can be enhanced
in frequency and spatial domains. This paper presents a
comparative review of some of the common methodologies in
frequency domain, that are used for satellite image resolution
enhancement.
Key Words: wavelet, transform,interpolation,DWT,bilinear,
bicubic, WZP, cycle-spinning.
1. INTRODUCTION
Satellite imaging is carried out by a high flying satellite or
aircraft that examines the earth to obtain knowledge about
the earth. There are a number of satellites that examine the
earth for different purposes. Satellite images are used in
plenty of applications such as medical images, agricultural
images, assisting plannersand engineers, extractingmineral
deposits, disaster mitigation planning and recovery. But,the
images captured by satellites suffer from a number of
artifacts due to very low frequencies. Hence in order to
upgrade the quality of images, there are a number of image
resolution enhancement methodologies designed and
implemented by different researchers.Thispaperrepresents
a study and comparison of different image resolution
enhancement strategies. The image resolutionenhancement
methodologies can be classified into two broad categories:
1.1 Frequency Domain methods
Frequency domain strategies work upon the Fourier
transform of the image. Fourier transform is computed for
the original image, all the processing operations are carried
out on this transform and then after all the operations, the
inverse of transform is applied to obtain a high resolution
image [6].
1.2 Spatial Domain Methods
Spatial domain imagesdirectly workuponthepixelsofthe
input images. The operations are directlycarriedonthepixel
values of the input image and finally a higher resolution
image is attained. But these techniques are not robust as
compared to frequency domain [10].
Image resolution enhancement infrequencydomainisbetter
as compared to spatial domain methods in all respects.
Frequency domain methodsmake use of Fourier transforms
and a technique called interpolation in order to enhance the
resolutionof the image. There are a number of interpolation
techniques that can be used to interject new pixels into an
image to enhance its quality [1][16].
2. Interpolation Based Frequency DomainMethods
Generally, interpolation is a technique that generates new
pixel points on the basis of the existing pixel values.
Interpolation is usedin conjunctionwithdifferenttransforms
in order to upgrade the resolution of satellite images. There
are different types of interpolation techniques that are used
in frequency domain [15].
2.1 Piecewise Linear Transformation: In this
methodology, each and every pixel point is considered for
manipulation. This enhances the resolution of an image by
considering mean dynamic range maximization. It was first
developed for monochrome images and then extended for
color images [11].
2.2 Nearest Neighbor Interpolation (NNI): Thisisthe
simplest form of interpolationthatdoesnot actuallyproduce
a new pixel. NNIworksby setting an emptypixelvalueinside
the image to a new value corresponding to the value of the
nearest pixel, that’s why the name nearest neighbor. This
results in enlargement of image pixels which causes some
irregularity in the edges of the images [7][8]. Thus NNI is the
least preferred technique for image resolutionenhancement.
2.3 Bilinear Interpolation: The issue of irregularity in
edges caused by nearest neighbor technique is solved by
bilinear interpolation. Bilinear interpolation considers four
diagonally nearest pixel values from the given pixel in order
to determine the relevant intensity of that pixel [12]. First of
all, 2X2 neighborhood of the unknown pixel is determined,
and then a weighted average of these is carried out in order
to find the appropriate value. This technique smoothens the
edges of the image but may cause a little blurriness in the
output image.
2.4 Bicubic Interpolation: In order to resolve the issues
of nearest neighbor interpolation and bilinear interpolation,
bicubic interpolation was designed. This strategy considers
the 4X4 neighborhood of the unknown pixel, taking into
account sixteen nearest pixels to determine the appropriate
pixel value of unknown point. As the range is wider, thus
bicubic interpolationproducesbetter results ascomparedto
nearest neighbor and bilinear interpolations.](https://image.slidesharecdn.com/irjet-v5i4330-190223083448/85/IRJET-A-Comparative-Review-of-Satellite-Image-Super-Resolution-Techniques-1-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1466
2.5 2D Wavelet Transformation: Wavelet
transformationis usedtodisintegrate the originalimageinto
a number of subband images with different frequencies.
Discretewavelet transform in twodimensionsproducesfour
subbandsnamely::LL(low-low),LH(low-high),HL(high-low),
and HH(high-high) subbands.The low-low subband is
considered approximate to the original input image [9][14].
Fig -1: 2D Wavelet Transformation
2.6 2D Wavelet Transformation and WZP: To carry
out the wavelet disintegration of an image, discrete wavelet
transform in one dimension is applied along the rows and
columnsof the image. Outof the subbandsproduced,thelow-
low subband is used as the input image for further
processing. Aftertransformation,waveletzeropadding(WZP)
is done. In WZP, matrices of zero are appended to the other
three subband images and then, inverse of discrete wavelet
transform is applied to obtain the output image [3]. WZP is a
simple technique, but results are not much better as
compared to those of bicubic interpolation.
Fig -2: 2D Wavelet Transformation and WZP
2.7 2D Wavelet Transformation and CS: In this
technique, first of all, an initial higher resolution image is
originated using WZP, then cycle spinning(CS) is carried out
where, a z-domain shift operator is applied on originated
image in horizontal and vertical directions. CS produces a
number of shifted images. After that, discrete wavelet
transform is applied and all the higher resolution subbands
are eliminated [13][16]. Finally, edge rectification is carried
out to obtain the final output image.
Fig -3: 2D Wavelet Transformation and CS
2.8 Complex WT: This technique is very similartowavelet
transformation. Likewise, complex wavelet transform
producestwolow frequency subbandsandsixhighfrequency
subbands, that have complex values (real and imaginary
parts). The higher frequency subbands are subjected to
bicubic interpolation by factor of α and a factor of α/2 to
attain a number of high frequency complex valued subbands
[10]. Finally, all these subband images are subjected to
inverse of CWT to produce the final high resolution output
image.
Fig -4 Complex WT
2.9 Discrete WT: This is a simple technique in which
wavelet transform is first of all applied to the original image
two produce the four subbands. Then, the higher frequency
subbandsare interpolated using bicubic interpolation which
produces high resolution coefficients, which are then
subjected to inverse of wavelet transform to attain the final
output image [4][5][17].
Fig -5 Discrete WT](https://image.slidesharecdn.com/irjet-v5i4330-190223083448/85/IRJET-A-Comparative-Review-of-Satellite-Image-Super-Resolution-Techniques-2-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1467
3. CONCLUSIONS
The review and study of various image resolution
enhancement techniques is done and it has been observed
that interpolation alone cannot upgrade theresolutionofthe
satellite images. Thus, wavelet transform in combination
with interpolation is used. Out of the existing interpolations,
bicubic interpolation is the best in terms of performance,
and thus, it is used with wavelet transforms to enhance the
image resolution [2][18].
REFERENCES
[1] A. Abirami, and N. Akshaya, “Enhancement of Satellite
Image Resolution with Moving Objects”, International
Journal of Electronicsand Communication Engineering,
Volume 4, Issue 6, pp. 22-27, 2013.
[2] A. Namdeo, and S. S. Bhadoriya, "A Review on Image
Enhancement Techniques with its Advantages and
Disadvantages", International Journal for Scientific
Research & Development, Volume 2, Issue 5, pp. 171-
182, 2016.
[3] A. R. Karande, and A. L. Renke, "A Review on Image
Enhancement Methods", International Journal of
Computer Applications, Volume 164, Issue 6, pp. 4-9,
2017.
[4] D. D. Buchade, and Prof. L.K. Chouthmol, "Image
Resolution Enhancement by using wavelet transform",
International Research Journal of Engineering and
Technology, Volume 4, Issue 6, pp. 1540-1544, 2017.
[5] H. Demirel, and G. Anbarjafari, “Image Resolution
Enhancement by using Discrete and Stationary Wavelet
Decomposition,” IEEE Transactions on Image
Processing, Volume 20, Issue 5, pp. 1458–1460, 2011.
[6] J. P. Bidishaw, and Dr. T. Nalini, "A Survey on Various
Image Enhancement Techniques", InternationalJournal
of Advanced Research in Computer Science, Volume 5,
Issue 3, pp. 160-162, 2014.
[7] M. A. Bendoumi, M. He, and S. Mei, “HyperspectralImage
Resolution Enhancement using High-Resolution
Multispectral Image based on Spectral Unmixing”, IEEE
transactions on Geosciences and Remote Sensing,
Volume 52, Issue 10, 2014.
[8] N. Tripathi, and A.P. K. G. Kirar, "Image Resolution
Enhancement by Wavelet Transform Based
Interpolation and Image Fusion", International Journal
of Advanced Research in Computer Science and
Software Engineering, Volume 4, Issue 8, pp. 318-323,
2014.
[9] O.Harikrishna, and A.Maheshwari “Satellite Image
Resolution Enhancement using DWT Technique”,
International Journal of SoftComputingandEngineering
Volume 2, Issue 5, 2012.
[10] P. Agrawal , V. Chourasia, R. Kapoor , and S. Agrawal, "A
Comprehensive Study of the Image Enhancement
Techniques", International Journal of Advance
Foundation and Research in Computer, Volume 1, Issue
7, pp. 84 -89, 2014.
[11] P. Janani, J. Premaladha, and K. S. Ravichandran, "Image
Enhancement Techniques: A Study", Indian Journal of
Science and Technology, Volume 8, Issue 22, pp. 1-12,
2015.
[12] P. Saranya, K. Fatimakani K, P. Kanchanadevi, S.
Venkatesan, and S. Govindaraju, "Survey on Wavelet
Domain Techniques for Image Super Resolution",
International Journal of Computer Science and Mobile
Computing, Volume 3, Issue 3, pp. 235-243, 2014.
[13] P.Suganya, N.Mohanapriya, and A.Vanitha “Survey on
Image Resolution Techniques for Satellite Images”
International Journal of Computer Science and
Information Technologies, Volume 4, Issue 6 ,pp. 835-
838, 2013.
[14] S. Azam, F. T. Zohra, and M. M. Islam, "A State-of-the-art
Review on Wavelet Based Image Resolution
Enhancement Techniques: Performance Evaluation
Criteria and Issues", International Journal of Image,
Graphics and Signal Processing, Volume 9, pp. 35-46,
2014.
[15] S. Dhariwal, "Comparative Analysis of Various Image
Enhancement Techniques", International Journal of
Electronics & Communication Technology, Volume 2,
Issue 3, pp. 91-95, 2011.
[16] S. Kaur, and P. Kaur, "Review and Analysis of Various
Image Enhancement Techniques", InternationalJournal
of Computer Applications Technology and Research,
Volume 4, Issue 5, pp. 414-418, 2015.
[17] S. Khidse, and M. Nagori, "Implementation and
Comparison of Image Enhancement Techniques",
International Journal of Computer Applications,Volume
96, Issue 4, pp. 9-16, 2014.
[18] W. Witwit, Y. Zhao, K. Jenkins, and Y. Zhao, "Satellite
image resolution enhancement using discrete wavelet
transform and new edge-directedinterpolation",Journal
of Electronic Imaging, Volume 26, Issue 2, 2017.](https://image.slidesharecdn.com/irjet-v5i4330-190223083448/85/IRJET-A-Comparative-Review-of-Satellite-Image-Super-Resolution-Techniques-3-320.jpg)
IRJET- A Comparative Review of Satellite Image Super Resolution Techniques
- 1. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072 © 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1465 A Comparative Review of Satellite Image Super resolution techniques Kangan Sardar1, Inderpreet Kaur2 1 M.Tech student, Computer science and Engineering, Bahra Group of Institutions, Patiala, India 2 Assistant Professor, Computer science and Engineering, Bahra Group of Institutions, Patiala, India ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract –Resolution of satellite imagesisaveryimportant factor and needs to be preserved for various applications. A number of techniques have been proposed by various researchers in this context. Image resolution can be enhanced in frequency and spatial domains. This paper presents a comparative review of some of the common methodologies in frequency domain, that are used for satellite image resolution enhancement. Key Words: wavelet, transform,interpolation,DWT,bilinear, bicubic, WZP, cycle-spinning. 1. INTRODUCTION Satellite imaging is carried out by a high flying satellite or aircraft that examines the earth to obtain knowledge about the earth. There are a number of satellites that examine the earth for different purposes. Satellite images are used in plenty of applications such as medical images, agricultural images, assisting plannersand engineers, extractingmineral deposits, disaster mitigation planning and recovery. But,the images captured by satellites suffer from a number of artifacts due to very low frequencies. Hence in order to upgrade the quality of images, there are a number of image resolution enhancement methodologies designed and implemented by different researchers.Thispaperrepresents a study and comparison of different image resolution enhancement strategies. The image resolutionenhancement methodologies can be classified into two broad categories: 1.1 Frequency Domain methods Frequency domain strategies work upon the Fourier transform of the image. Fourier transform is computed for the original image, all the processing operations are carried out on this transform and then after all the operations, the inverse of transform is applied to obtain a high resolution image [6]. 1.2 Spatial Domain Methods Spatial domain imagesdirectly workuponthepixelsofthe input images. The operations are directlycarriedonthepixel values of the input image and finally a higher resolution image is attained. But these techniques are not robust as compared to frequency domain [10]. Image resolution enhancement infrequencydomainisbetter as compared to spatial domain methods in all respects. Frequency domain methodsmake use of Fourier transforms and a technique called interpolation in order to enhance the resolutionof the image. There are a number of interpolation techniques that can be used to interject new pixels into an image to enhance its quality [1][16]. 2. Interpolation Based Frequency DomainMethods Generally, interpolation is a technique that generates new pixel points on the basis of the existing pixel values. Interpolation is usedin conjunctionwithdifferenttransforms in order to upgrade the resolution of satellite images. There are different types of interpolation techniques that are used in frequency domain [15]. 2.1 Piecewise Linear Transformation: In this methodology, each and every pixel point is considered for manipulation. This enhances the resolution of an image by considering mean dynamic range maximization. It was first developed for monochrome images and then extended for color images [11]. 2.2 Nearest Neighbor Interpolation (NNI): Thisisthe simplest form of interpolationthatdoesnot actuallyproduce a new pixel. NNIworksby setting an emptypixelvalueinside the image to a new value corresponding to the value of the nearest pixel, that’s why the name nearest neighbor. This results in enlargement of image pixels which causes some irregularity in the edges of the images [7][8]. Thus NNI is the least preferred technique for image resolutionenhancement. 2.3 Bilinear Interpolation: The issue of irregularity in edges caused by nearest neighbor technique is solved by bilinear interpolation. Bilinear interpolation considers four diagonally nearest pixel values from the given pixel in order to determine the relevant intensity of that pixel [12]. First of all, 2X2 neighborhood of the unknown pixel is determined, and then a weighted average of these is carried out in order to find the appropriate value. This technique smoothens the edges of the image but may cause a little blurriness in the output image. 2.4 Bicubic Interpolation: In order to resolve the issues of nearest neighbor interpolation and bilinear interpolation, bicubic interpolation was designed. This strategy considers the 4X4 neighborhood of the unknown pixel, taking into account sixteen nearest pixels to determine the appropriate pixel value of unknown point. As the range is wider, thus bicubic interpolationproducesbetter results ascomparedto nearest neighbor and bilinear interpolations.
- 2. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072 © 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1466 2.5 2D Wavelet Transformation: Wavelet transformationis usedtodisintegrate the originalimageinto a number of subband images with different frequencies. Discretewavelet transform in twodimensionsproducesfour subbandsnamely::LL(low-low),LH(low-high),HL(high-low), and HH(high-high) subbands.The low-low subband is considered approximate to the original input image [9][14]. Fig -1: 2D Wavelet Transformation 2.6 2D Wavelet Transformation and WZP: To carry out the wavelet disintegration of an image, discrete wavelet transform in one dimension is applied along the rows and columnsof the image. Outof the subbandsproduced,thelow- low subband is used as the input image for further processing. Aftertransformation,waveletzeropadding(WZP) is done. In WZP, matrices of zero are appended to the other three subband images and then, inverse of discrete wavelet transform is applied to obtain the output image [3]. WZP is a simple technique, but results are not much better as compared to those of bicubic interpolation. Fig -2: 2D Wavelet Transformation and WZP 2.7 2D Wavelet Transformation and CS: In this technique, first of all, an initial higher resolution image is originated using WZP, then cycle spinning(CS) is carried out where, a z-domain shift operator is applied on originated image in horizontal and vertical directions. CS produces a number of shifted images. After that, discrete wavelet transform is applied and all the higher resolution subbands are eliminated [13][16]. Finally, edge rectification is carried out to obtain the final output image. Fig -3: 2D Wavelet Transformation and CS 2.8 Complex WT: This technique is very similartowavelet transformation. Likewise, complex wavelet transform producestwolow frequency subbandsandsixhighfrequency subbands, that have complex values (real and imaginary parts). The higher frequency subbands are subjected to bicubic interpolation by factor of α and a factor of α/2 to attain a number of high frequency complex valued subbands [10]. Finally, all these subband images are subjected to inverse of CWT to produce the final high resolution output image. Fig -4 Complex WT 2.9 Discrete WT: This is a simple technique in which wavelet transform is first of all applied to the original image two produce the four subbands. Then, the higher frequency subbandsare interpolated using bicubic interpolation which produces high resolution coefficients, which are then subjected to inverse of wavelet transform to attain the final output image [4][5][17]. Fig -5 Discrete WT
- 3. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072 © 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1467 3. CONCLUSIONS The review and study of various image resolution enhancement techniques is done and it has been observed that interpolation alone cannot upgrade theresolutionofthe satellite images. Thus, wavelet transform in combination with interpolation is used. Out of the existing interpolations, bicubic interpolation is the best in terms of performance, and thus, it is used with wavelet transforms to enhance the image resolution [2][18]. REFERENCES [1] A. Abirami, and N. Akshaya, “Enhancement of Satellite Image Resolution with Moving Objects”, International Journal of Electronicsand Communication Engineering, Volume 4, Issue 6, pp. 22-27, 2013. [2] A. Namdeo, and S. S. Bhadoriya, "A Review on Image Enhancement Techniques with its Advantages and Disadvantages", International Journal for Scientific Research & Development, Volume 2, Issue 5, pp. 171- 182, 2016. [3] A. R. Karande, and A. L. Renke, "A Review on Image Enhancement Methods", International Journal of Computer Applications, Volume 164, Issue 6, pp. 4-9, 2017. [4] D. D. Buchade, and Prof. L.K. Chouthmol, "Image Resolution Enhancement by using wavelet transform", International Research Journal of Engineering and Technology, Volume 4, Issue 6, pp. 1540-1544, 2017. [5] H. Demirel, and G. Anbarjafari, “Image Resolution Enhancement by using Discrete and Stationary Wavelet Decomposition,” IEEE Transactions on Image Processing, Volume 20, Issue 5, pp. 1458–1460, 2011. [6] J. P. Bidishaw, and Dr. T. Nalini, "A Survey on Various Image Enhancement Techniques", InternationalJournal of Advanced Research in Computer Science, Volume 5, Issue 3, pp. 160-162, 2014. [7] M. A. Bendoumi, M. He, and S. Mei, “HyperspectralImage Resolution Enhancement using High-Resolution Multispectral Image based on Spectral Unmixing”, IEEE transactions on Geosciences and Remote Sensing, Volume 52, Issue 10, 2014. [8] N. Tripathi, and A.P. K. G. Kirar, "Image Resolution Enhancement by Wavelet Transform Based Interpolation and Image Fusion", International Journal of Advanced Research in Computer Science and Software Engineering, Volume 4, Issue 8, pp. 318-323, 2014. [9] O.Harikrishna, and A.Maheshwari “Satellite Image Resolution Enhancement using DWT Technique”, International Journal of SoftComputingandEngineering Volume 2, Issue 5, 2012. [10] P. Agrawal , V. Chourasia, R. Kapoor , and S. Agrawal, "A Comprehensive Study of the Image Enhancement Techniques", International Journal of Advance Foundation and Research in Computer, Volume 1, Issue 7, pp. 84 -89, 2014. [11] P. Janani, J. Premaladha, and K. S. Ravichandran, "Image Enhancement Techniques: A Study", Indian Journal of Science and Technology, Volume 8, Issue 22, pp. 1-12, 2015. [12] P. Saranya, K. Fatimakani K, P. Kanchanadevi, S. Venkatesan, and S. Govindaraju, "Survey on Wavelet Domain Techniques for Image Super Resolution", International Journal of Computer Science and Mobile Computing, Volume 3, Issue 3, pp. 235-243, 2014. [13] P.Suganya, N.Mohanapriya, and A.Vanitha “Survey on Image Resolution Techniques for Satellite Images” International Journal of Computer Science and Information Technologies, Volume 4, Issue 6 ,pp. 835- 838, 2013. [14] S. Azam, F. T. Zohra, and M. M. Islam, "A State-of-the-art Review on Wavelet Based Image Resolution Enhancement Techniques: Performance Evaluation Criteria and Issues", International Journal of Image, Graphics and Signal Processing, Volume 9, pp. 35-46, 2014. [15] S. Dhariwal, "Comparative Analysis of Various Image Enhancement Techniques", International Journal of Electronics & Communication Technology, Volume 2, Issue 3, pp. 91-95, 2011. [16] S. Kaur, and P. Kaur, "Review and Analysis of Various Image Enhancement Techniques", InternationalJournal of Computer Applications Technology and Research, Volume 4, Issue 5, pp. 414-418, 2015. [17] S. Khidse, and M. Nagori, "Implementation and Comparison of Image Enhancement Techniques", International Journal of Computer Applications,Volume 96, Issue 4, pp. 9-16, 2014. [18] W. Witwit, Y. Zhao, K. Jenkins, and Y. Zhao, "Satellite image resolution enhancement using discrete wavelet transform and new edge-directedinterpolation",Journal of Electronic Imaging, Volume 26, Issue 2, 2017.