Digital image compression is a technique that
allows to reduce the size of an image in order to increase the
capacity storage devices and to optimize the use of network
bandwidth. The quality of compressed images with the
techniques based on the discrete cosine transform or the
wavelet transform is generally measured with PSNR or
SSIM. Theses metrics are not suitable to images compressed
with the singular values decomposition. This paper presents
a new metric based on the energy ratio to measure the
quality of the images coded with the SVD. A series of tests
on 512 × 512 pixels images show that, for a rank k = 40
corresponding to a SSIM = 0,94 or PSNR = 35 dB, 99,9% of
the energy are restored. Three areas of image quality
assessments were identified. This new metric is also very
accurate and could overcome the weaknesses of PSNR and
SSIM.
Published In:IJCSN Journal Volume 5, Issue 6
Date of Publication : December 2016
Pages : 960-965
Figures :06
Tables : --
Henri Bruno Razafindradina : was born in Fianarantsoa,
Madagascar, on 1978. He received, respectively, his M.S degree and
Ph.D in Computer Science and Information Engineering in 2005 and
2008. He served since 2010 as a professor at Higher Institute of
Technology Diego Suarez, became an assistant lecturer in 2011. His
current research interests include images compression, multimedia,
computer vision, information hiding.
Nicolas Raft Razafindrakoto : is a professor at the Higher
Polytechnic School of Antananarivo. His current research interests
include petri networks and computer science.
Paul Auguste Randriamitantsoa : is a professor at the Higher
Polytechnic School of Antananarivo. His current research interests
include automatic and computer science.
Energy Radio can be used to complete the weaknesses of
PSNR and SSIM that are only visual indicators. The
second area coincides well with the threshold of
appreciation of the PSNR which says that between 30 dB
and 50 dB, the image is good quality. But the gap from 5
dB indicates that with the algorithm SVD, the typical
value of the PSNR is 35 dB. As the difference between
Emin and Emax is lower than 0,1; we can say that the
metrics is very accurate with regard to SSIM and PSNR.
The energy ratio is a simple indicator to use but require
an important calculation time. It could be improved by
using the parallel algorithm [8] proposed by Saira Banu.
The next step will be to apply the energy ratio to the color
images. We could also use it to evaluate the hybrid
algorithms such as SVD and MPQ-BTC [9].
[1] A. B. Watson, « Image Compression Using the
Discrete Cosine Transform », Mathematica Journal,
1994, pp. 81-88.
[2] S. C. Meadows, « Color image compression using
wavelet transform », Thesis in Electrical Engineering,
1997, pp. 1-86.
[3] J. Chen, « Image compression with SVD », ECS 289K
Scientific Computation, 2000, pp. 13
[4] A. Abrahamsen and D. Richards, « Image
Compression using Singular Value Decomposition »,
Linear algebra applications, 2001, pp. 1-14.
[5] S. Kahu, R. Rahate, « Image Compression using
Singular Value Decomposition », International Journal
of Advancements in Research & Technology, 2013,
Volume 2, Issue 8, pp. 244-248.
[6] A. R. Sadek, « SVD Based Images Processing
Applications : State of the Art, Contributions and
Research Challenges », International Journal of
Advanced Computer Science and Applications, 2012,
Volume 3, No. 7, pp. 26-34.
[7] Z. Wang, A. C. Bovik, H. R. Sheikh and E. P.
Simoncelli, « Image quality assessment :From error
visibility to structural similarity », IEEE Transactions
on Image Processing, 2004, Volume 13, no. 4, pp.
600-612.
[8] J. SairaBanu, B. Rajasekhara and R. Pandey,
« Parallel Implementation of Singular Value
Decomposition in Image Compression Using Open Mp
and Sparse Matrix Representation », Indian Journal
of Science and Technology, 2015, Volume 8 (13), pp.
1-10.
[9] N. K. El Abbadi and Al, « Image Compression based
on SVD and MPQ-BTC », Journal of Computer
Science, 2014, Volume 10, pp. 2095-2104.