Image Resolution Enhancement using DWT and Spatial Domain Interpolation Technique
0 likes52 views
Image Resolution is one of the important quality metrics of images. Images with high resolution are required in many fields. In this paper, a new resolution enhancement technique is proposed based on the interpolation of four sub band images generated by Discrete Wavelet Transform (DWT) and the original Low Resolution (LR) input image. In this technique, the four sub band images generated by DWT and the input LR image are interpolated with scaling factor, α and then performed inverse DWT to obtain the intermediate High Resolution (HR) Image. The difference between the intermediate HR image and the interpolated LR input image is added to the intermediate HR image to obtain final output HR Image. Lanczos interpolation is used in this technique. The proposed technique is tested on well known bench mark images. The quantitative and visual results shows the superiority of the proposed technique over the conventional and state of art image resolution enhancement techniques in wavelet domain using haar wavelet filter.
![Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96
www.ijera.com 93|P a g e
Image Resolution Enhancement using DWT and Spatial Domain
Interpolation Technique
Mrs. G. Padma Priya*, Prof. T. Venkateswarlu**
*(Research Scholar, Department of ECE, SVUCE, SV University, Tirupati.)
** (Professor, Department of ECE, SVUCE, SV University, Tirupati.)
ABSTRACT
Image Resolution is one of the important quality metrics of images. Images with high resolution are required in
many fields. In this paper, a new resolution enhancement technique is proposed based on the interpolation of
four sub band images generated by Discrete Wavelet Transform (DWT) and the original Low Resolution (LR)
input image. In this technique, the four sub band images generated by DWT and the input LR image are
interpolated with scaling factor, α and then performed inverse DWT to obtain the intermediate High Resolution
(HR) Image. The difference between the intermediate HR image and the interpolated LR input image is added
to the intermediate HR image to obtain final output HR Image. Lanczos interpolation is used in this technique.
The proposed technique is tested on well known bench mark images. The quantitative and visual results shows
the superiority of the proposed technique over the conventional and state of art image resolution enhancement
techniques in wavelet domain using haar wavelet filter.
Keywords: DWT, Lanczos Interpolation, Resolution.
I. INTRODUCTION
Image Resolution is one of the most important
quality metrics of images and videos. Images with
higher resolution are required in most of the imaging
applications, such as, medical imaging, video
standard conversion, remote sensing and surveillance
video. Resolution of an image stands for number of
pixels in image. Image with more number of pixels
has high resolution. The pixel resolution can be
specified with the set of two positive integer
numbers, where the first number is the number of
pixel columns (width) and the second is the number
of pixel rows (height), for example as 512 x 512. The
most widely used technique for enhancing the image
resolution is Interpolation. Fundamentally,
Interpolation is the process of using known data to
estimate values at unknown locations [1]. In Image
processing, Interpolation is a method to increase the
number of pixels in digital image. Conventional
Interpolation Techniques which are commonly used
are Nearest Neighbor, Bilinear, Bicubic and Lanczos.
Resolution Enhancement techniques which are not
based on wavelets suffer from the drawback of losing
high frequency contents which results in blurring of
the images [2]. Recently some techniques have been
proposed [2]-[7] in wavelet domain for resolution
enhancement. Using Wavelet Transform, spectrum
can be obtained as a function of shift and scale.
Hence, it is suitable for obtaining spatial as well as
spectral resolution enhancement.
By using DWT, a HR Image can be decomposed
into a LR Image and three wavelet detail images with
horizontal, vertical and diagonal edge information at
each scale by applying the 1D - DWT along the rows
of the image first, and then the 1D - DWT along the
column of the image. These four sub band images are
referred to as LL, LH, HL, HH sub bands. The
frequency components of these sub bands cover full
frequency spectrum of the original image. Inverse
DWT is used to obtain the original image using these
four sub bands. The block diagram, representing the
2D – DWT process was given in Fig.1 and the
corresponding output images for single level
decomposition was given in Fig.2.
(a)
(b)
Fig. 1: (a) Single level decomposition of 2D DWT
(b) Single level 2D Inverse DWT.
RESEARCH ARTICLE OPEN ACCESS](https://image.slidesharecdn.com/p62039396-160727072010/85/Image-Resolution-Enhancement-using-DWT-and-Spatial-Domain-Interpolation-Technique-1-320.jpg)
![Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96
www.ijera.com 94|P a g e
Fig. 2: Single level decomposition of lena image.
II. PROPOSED TECHNIQUE
In the proposed technique, the input LR image is
decomposed into four sub band images using DWT
and interpolated with lanczos kernel with scaling
factor, α. Interpolated sub band images of LL, LH,
HL and HH sub band images are represented with
ILL, ILH, IHL and IHH respectively. Intermediate
HR image is obtained by applying inverse DWT,
with the interpolated sub band images ILL, ILH, IHL
and IHH. The intermediate HR image is subtracted
from interpolated input image with scaling factor, α
to obtain the difference image, which is added to the
intermediate HR image to get final output HR image.
The input LR image was generated using two
consecutive decomposition of original HR image
using DWT with haar wavelet. The algorithm of the
proposed technique is followed from Fig.3.
Fig.3: Proposed Image resolution enhancement
technique using DWT and lanczos3 interpolation.
III. PERFORMANCE EVALUATION
CRITERIA AND IMAGE QUALITY
MEASURE
The resolution of the test image used in
evaluation of image interpolation technique is known.
After interpolation this resolution will change. To
evaluate picture quality, the interpolated image will
be compared with the original input image. In these
circumstances input image and interpolated
image cannot be compared because of different
resolutions. Common approach is to start with an
original HR image, generate a lower resolution
version of original image by downscaling, and then
use different interpolation methods to upscale low
resolution image [8]. After that original and
magnified HR images are compared to evaluate
different techniques using different picture quality
measures. Peak Signal to Noise Ratio (PSNR) is used
for comparing different image resolution
enhancement techniques.
PSNR is the ratio between the maximum
possible power of a signal and the power of noise.
PSNR is usually expressed in terms of the
logarithmic decibel scale and they can be expressed
as:
10
2
10log ( )R
MSE
PSNR (1)
Where
R is the maximum fluctuation in the input image
(R=255, if images are represented with 8-bit gray](https://image.slidesharecdn.com/p62039396-160727072010/85/Image-Resolution-Enhancement-using-DWT-and-Spatial-Domain-Interpolation-Technique-2-320.jpg)
![Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96
www.ijera.com 95|P a g e
scale representation with radiometric resolution
of 8-bit)
MSE represents the Mean Square Error between
the given original image, a and interpolated
image, b with size M x N and is given by the
formula
2
, ,
,
( )i j i j
i j
X
a b
MSE
M N
(2)
IV. RESULTS
The proposed technique was implemented using
MATLAB R2014a software. The performance of
proposed technique was compared with bicubic
interpolation technique and wavelet based resolution
enhancement techniques such as wavelet zero
padding (WZP)[3], DASR method with bicubic
interpolation and haar wavelet[5]. The proposed
method gives better performance with lanczos3
interpolation and haar wavelet for unsigned 8 bit
integer images. In Table 1, PSNR is compared for the
existing and proposed method. The results
demonstrate the superiority of the proposed technique
over the above specified techniques with haar
wavelet.
Table1: PSNR (in dB)for images, up scaled from
128x128 to 512x512 with scaling factor of 4.
(a) (b)
(c) (d)
(e) (f)
Fig. 4: Lena image upscaled from 128x128 to
512x512 with scaling factor of 4. (a) Original HR
image (b) Generated input LR image, HR output
image and the difference image from left to right, (c)
and (d) of bicubic method, (e) and (f) of proposed
technique.
(a) (b)
(c) (d)
(e) (f)
Fig. 5: Elaine image upscaled from 128x128 to
512x512 with scaling factor of 4. (a) Original HR
image (b) Generated input LR image, HR output
image and the difference image from left to right, (c)
and (d) of bicubic method, (e) and (f) of proposed
technique.
(a) (b)
Method
Image
PSNR (in dB)
Lena Elaine Baboon
Bicubic 26.86 28.93 20.61
WZP (Haar) [3] 26.67 28.06 21.11
DASR
(bicubic+haar) [5]
27.07 27.94 18.06
DWT & SWT RE[6] 34.82 35.01 23.87
Proposed Method
(lanczos3+haar)
34.87 42.12 24.81](https://image.slidesharecdn.com/p62039396-160727072010/85/Image-Resolution-Enhancement-using-DWT-and-Spatial-Domain-Interpolation-Technique-3-320.jpg)
![Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96
www.ijera.com 96|P a g e
(c) (d)
Fig. 6: Baboon image upscaled from 128x128 to
512x512 with scaling factor of 4. (a) Original HR
image (b) Generated input LR image, (c) HR output
image and (d) The difference image of proposed
technique.
The original HR image (512x512), input LR
image (128x128), output HR image and difference
image obtained by using bicubic and proposed
technique are shown in Fig.4 and Fig.5. The
subjective results show that the proposed method has
less difference when compared with the above
specified methods.
V. CONCLUSION
In this paper, a new technique for image
resolution enhancement using DWT with haar
wavelet and lanczos3 interpolation technique is
presented. Performance evaluation criteria and image
quality measure, PSNR is discussed in this paper.
The PSNR value and visual results demonstrate the
superiority of the proposed technique over DASR
method with haar wavelet and bicubic interpolation.
REFERENCES
[1] R.C. Gonzalez and R.E. Woods, Digital
Image Processing: 3rd
edition, Pearson
Education Inc. ©2008.
[2] M. Z. Iqbal, A. Ghafoor and A. M. Siddiqui.,
“Satellite Image Resolution Enhancement
Using Dual – Tree Complex Wavelet
Transform and Nonlocal Means”, IEEE
Geoscience and remote sensing Letters, Vol.
10, No.3, May. 2013, pp. 451-455.
[3] A. Temizel and T. Vlachos, “Wavelet
Domain Image Resolution Enhancement
Using Cycle Spinning and Edge Modelling”,
13th
European Signal Processing conference,
Sep. 2005, pp.203-205.
[4] H. Demirel and G. Anbarjafari, “Satellite
Image Resolution Enhancement Using
Complex Wavelet Transform”, IEEE
Geoscience and remote sensing Letters, Vol.
7, No.1, Jan. 2010, pp. 123-126.
[5] G. Anbarjafari and H. Demirel, “Image Super
Resolution Based on Interpolation of Wavelet
Domain High Frequency Subbands and the
Spatial Domain Input Image”, ETRI Journal,
Vol. 32, No.3, Jun. 2010, pp. 390-394.
[6] H. Demirel and G. Anbarjafari, “IMAGE
Resolution Enhancement by Using Discrete
and Stationary Wavelet Decomposition”,
IEEE Transactions on Image Processing, Vol.
20, No.5, May 2011, pp. 1458-1460.
[7] H. Demirel and G. Anbarjafari, “Discrete
Wavelet Transform – Based Satellite Image
Resolution Enhancement”, IEEE
Transactions on Geoscience and remote
sensing, Vol. 49, No.6, June 2011, pp. 1997-
2004.
[8] D. Emil, G. Sonja, G. Mislav, “The Use of
Wavelets in Image Interpolation: Possibilities
and Limitations”, RADIOENGINEERING,
Vol. 16, No .4, Dec. 2007, pp. 101-109.](https://image.slidesharecdn.com/p62039396-160727072010/85/Image-Resolution-Enhancement-using-DWT-and-Spatial-Domain-Interpolation-Technique-4-320.jpg)
Ad
More Related Content
What's hot (18)
Similar to Image Resolution Enhancement using DWT and Spatial Domain Interpolation Technique (20)
Ad
Recently uploaded (20)
Ad
Image Resolution Enhancement using DWT and Spatial Domain Interpolation Technique
- 1. Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96 www.ijera.com 93|P a g e Image Resolution Enhancement using DWT and Spatial Domain Interpolation Technique Mrs. G. Padma Priya*, Prof. T. Venkateswarlu** *(Research Scholar, Department of ECE, SVUCE, SV University, Tirupati.) ** (Professor, Department of ECE, SVUCE, SV University, Tirupati.) ABSTRACT Image Resolution is one of the important quality metrics of images. Images with high resolution are required in many fields. In this paper, a new resolution enhancement technique is proposed based on the interpolation of four sub band images generated by Discrete Wavelet Transform (DWT) and the original Low Resolution (LR) input image. In this technique, the four sub band images generated by DWT and the input LR image are interpolated with scaling factor, α and then performed inverse DWT to obtain the intermediate High Resolution (HR) Image. The difference between the intermediate HR image and the interpolated LR input image is added to the intermediate HR image to obtain final output HR Image. Lanczos interpolation is used in this technique. The proposed technique is tested on well known bench mark images. The quantitative and visual results shows the superiority of the proposed technique over the conventional and state of art image resolution enhancement techniques in wavelet domain using haar wavelet filter. Keywords: DWT, Lanczos Interpolation, Resolution. I. INTRODUCTION Image Resolution is one of the most important quality metrics of images and videos. Images with higher resolution are required in most of the imaging applications, such as, medical imaging, video standard conversion, remote sensing and surveillance video. Resolution of an image stands for number of pixels in image. Image with more number of pixels has high resolution. The pixel resolution can be specified with the set of two positive integer numbers, where the first number is the number of pixel columns (width) and the second is the number of pixel rows (height), for example as 512 x 512. The most widely used technique for enhancing the image resolution is Interpolation. Fundamentally, Interpolation is the process of using known data to estimate values at unknown locations [1]. In Image processing, Interpolation is a method to increase the number of pixels in digital image. Conventional Interpolation Techniques which are commonly used are Nearest Neighbor, Bilinear, Bicubic and Lanczos. Resolution Enhancement techniques which are not based on wavelets suffer from the drawback of losing high frequency contents which results in blurring of the images [2]. Recently some techniques have been proposed [2]-[7] in wavelet domain for resolution enhancement. Using Wavelet Transform, spectrum can be obtained as a function of shift and scale. Hence, it is suitable for obtaining spatial as well as spectral resolution enhancement. By using DWT, a HR Image can be decomposed into a LR Image and three wavelet detail images with horizontal, vertical and diagonal edge information at each scale by applying the 1D - DWT along the rows of the image first, and then the 1D - DWT along the column of the image. These four sub band images are referred to as LL, LH, HL, HH sub bands. The frequency components of these sub bands cover full frequency spectrum of the original image. Inverse DWT is used to obtain the original image using these four sub bands. The block diagram, representing the 2D – DWT process was given in Fig.1 and the corresponding output images for single level decomposition was given in Fig.2. (a) (b) Fig. 1: (a) Single level decomposition of 2D DWT (b) Single level 2D Inverse DWT. RESEARCH ARTICLE OPEN ACCESS
- 2. Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96 www.ijera.com 94|P a g e Fig. 2: Single level decomposition of lena image. II. PROPOSED TECHNIQUE In the proposed technique, the input LR image is decomposed into four sub band images using DWT and interpolated with lanczos kernel with scaling factor, α. Interpolated sub band images of LL, LH, HL and HH sub band images are represented with ILL, ILH, IHL and IHH respectively. Intermediate HR image is obtained by applying inverse DWT, with the interpolated sub band images ILL, ILH, IHL and IHH. The intermediate HR image is subtracted from interpolated input image with scaling factor, α to obtain the difference image, which is added to the intermediate HR image to get final output HR image. The input LR image was generated using two consecutive decomposition of original HR image using DWT with haar wavelet. The algorithm of the proposed technique is followed from Fig.3. Fig.3: Proposed Image resolution enhancement technique using DWT and lanczos3 interpolation. III. PERFORMANCE EVALUATION CRITERIA AND IMAGE QUALITY MEASURE The resolution of the test image used in evaluation of image interpolation technique is known. After interpolation this resolution will change. To evaluate picture quality, the interpolated image will be compared with the original input image. In these circumstances input image and interpolated image cannot be compared because of different resolutions. Common approach is to start with an original HR image, generate a lower resolution version of original image by downscaling, and then use different interpolation methods to upscale low resolution image [8]. After that original and magnified HR images are compared to evaluate different techniques using different picture quality measures. Peak Signal to Noise Ratio (PSNR) is used for comparing different image resolution enhancement techniques. PSNR is the ratio between the maximum possible power of a signal and the power of noise. PSNR is usually expressed in terms of the logarithmic decibel scale and they can be expressed as: 10 2 10log ( )R MSE PSNR (1) Where R is the maximum fluctuation in the input image (R=255, if images are represented with 8-bit gray
- 3. Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96 www.ijera.com 95|P a g e scale representation with radiometric resolution of 8-bit) MSE represents the Mean Square Error between the given original image, a and interpolated image, b with size M x N and is given by the formula 2 , , , ( )i j i j i j X a b MSE M N (2) IV. RESULTS The proposed technique was implemented using MATLAB R2014a software. The performance of proposed technique was compared with bicubic interpolation technique and wavelet based resolution enhancement techniques such as wavelet zero padding (WZP)[3], DASR method with bicubic interpolation and haar wavelet[5]. The proposed method gives better performance with lanczos3 interpolation and haar wavelet for unsigned 8 bit integer images. In Table 1, PSNR is compared for the existing and proposed method. The results demonstrate the superiority of the proposed technique over the above specified techniques with haar wavelet. Table1: PSNR (in dB)for images, up scaled from 128x128 to 512x512 with scaling factor of 4. (a) (b) (c) (d) (e) (f) Fig. 4: Lena image upscaled from 128x128 to 512x512 with scaling factor of 4. (a) Original HR image (b) Generated input LR image, HR output image and the difference image from left to right, (c) and (d) of bicubic method, (e) and (f) of proposed technique. (a) (b) (c) (d) (e) (f) Fig. 5: Elaine image upscaled from 128x128 to 512x512 with scaling factor of 4. (a) Original HR image (b) Generated input LR image, HR output image and the difference image from left to right, (c) and (d) of bicubic method, (e) and (f) of proposed technique. (a) (b) Method Image PSNR (in dB) Lena Elaine Baboon Bicubic 26.86 28.93 20.61 WZP (Haar) [3] 26.67 28.06 21.11 DASR (bicubic+haar) [5] 27.07 27.94 18.06 DWT & SWT RE[6] 34.82 35.01 23.87 Proposed Method (lanczos3+haar) 34.87 42.12 24.81
- 4. Mrs. G. Padma Priya Int. Journal of Engineering Research and Applications www.ijera.com ISSN: 2248-9622, Vol. 6, Issue 2, (Part - 3) February 2016, pp.93-96 www.ijera.com 96|P a g e (c) (d) Fig. 6: Baboon image upscaled from 128x128 to 512x512 with scaling factor of 4. (a) Original HR image (b) Generated input LR image, (c) HR output image and (d) The difference image of proposed technique. The original HR image (512x512), input LR image (128x128), output HR image and difference image obtained by using bicubic and proposed technique are shown in Fig.4 and Fig.5. The subjective results show that the proposed method has less difference when compared with the above specified methods. V. CONCLUSION In this paper, a new technique for image resolution enhancement using DWT with haar wavelet and lanczos3 interpolation technique is presented. Performance evaluation criteria and image quality measure, PSNR is discussed in this paper. The PSNR value and visual results demonstrate the superiority of the proposed technique over DASR method with haar wavelet and bicubic interpolation. REFERENCES [1] R.C. Gonzalez and R.E. Woods, Digital Image Processing: 3rd edition, Pearson Education Inc. ©2008. [2] M. Z. Iqbal, A. Ghafoor and A. M. Siddiqui., “Satellite Image Resolution Enhancement Using Dual – Tree Complex Wavelet Transform and Nonlocal Means”, IEEE Geoscience and remote sensing Letters, Vol. 10, No.3, May. 2013, pp. 451-455. [3] A. Temizel and T. Vlachos, “Wavelet Domain Image Resolution Enhancement Using Cycle Spinning and Edge Modelling”, 13th European Signal Processing conference, Sep. 2005, pp.203-205. [4] H. Demirel and G. Anbarjafari, “Satellite Image Resolution Enhancement Using Complex Wavelet Transform”, IEEE Geoscience and remote sensing Letters, Vol. 7, No.1, Jan. 2010, pp. 123-126. [5] G. Anbarjafari and H. Demirel, “Image Super Resolution Based on Interpolation of Wavelet Domain High Frequency Subbands and the Spatial Domain Input Image”, ETRI Journal, Vol. 32, No.3, Jun. 2010, pp. 390-394. [6] H. Demirel and G. Anbarjafari, “IMAGE Resolution Enhancement by Using Discrete and Stationary Wavelet Decomposition”, IEEE Transactions on Image Processing, Vol. 20, No.5, May 2011, pp. 1458-1460. [7] H. Demirel and G. Anbarjafari, “Discrete Wavelet Transform – Based Satellite Image Resolution Enhancement”, IEEE Transactions on Geoscience and remote sensing, Vol. 49, No.6, June 2011, pp. 1997- 2004. [8] D. Emil, G. Sonja, G. Mislav, “The Use of Wavelets in Image Interpolation: Possibilities and Limitations”, RADIOENGINEERING, Vol. 16, No .4, Dec. 2007, pp. 101-109.