Sunteți pe pagina 1din 4

A Comparative Analysis between SPIHT and JPEG2000

Compression Methods
HUSAM ABDULQUDOS AHMED ALWARETH, M021610016
Faculty of Electronics and Computer Engineering
University Technical Malaysia Melaka (UTeM)
e-mail: hussam.w@hotmail.com

Abstract The objective of this paper is to compare and particular study exact relating of the rate of decomposition in
analysis the quality of images after being compressed. There are the error between the compressed image and the original image
many techniques used for compression images such as Karhunen- [2][3].
Loeve Transform (KLT), Discrete Fourier Transform (DFT),
Discrete Sine Transform (DST), Walsh Hadamard Transform
In (1993) David Jeff Jackson classify the term of the image data
(WHT), Discrete Cosine Transform (DCT), and Discrete Wavelet
Transform (DWT). Whereas, in this task performed image compression as it is usable to the image processing. By the
compression using modules Set-Partitioning in Hierarchical Trees Where an analysis of numerous images compression scenarios
(SPIHT) and JPEG-2000 Standard (JPEG2000).The compression are discussed with their relative performance. Many topics
conducted using Vcdemo software with varying the levels of bit- focused on the image compression are tested during study such
rates. The result values of Mean Square Error (MSE), signal to as the generic data of image compression algorithms, the file
Noise Ratio (SNR), and Peak Signal to Noise Ratio (PSNR) that format schemes. While, in 1997 Maire D. Reavy et.al shows a
determine the quality of the compressed image of each module are new technique of the lossless of bi level for the image
presented. From the result obtained, the method of SPIHT is compression which represented to substitute JBIG and G3, the
better for compressing images due to high PSNR
recent standards for the bi level and for the image compression.
Index Terms JPEG2000; SPIHT; MSE; SNR; PSNR. In 2000 Mitchell A. Golner et.al explained the concepts behind
optimized of the image compression ratio by using the data
I. INTRODUCTION related to the signals characteristic and their applications. In
The huge usage of digital multimedia via communications, addition, in 2003 Deepti Gupta et.al Illustrates that the
wireless communications, Internet, Intranet and cellular mobile quantization and the wavelet transform techniques have
leads to incurable growth of data flow through these Media. The generate the algorithm able of surpassing the current image
researchers go deep in developing efficient techniques in these compression standards. By the 2007 Li Ke et.al presented that
fields such as compression of data, image and video. Recently, for the wavelet transform when is use for the image
video compression techniques and their applications in many compression, the selected wavelet fundamental influence the
areas (educational, agriculture, medical). In contrast, recently efficiency of the coding reconstructed image as well as the
numerous technologies are developed and rapidly spread quality, due to the characteristics of the parameters for diverse
worldwide for the purpose of compression as well as the wavelet fundamental which are diverse. In 2010 a comparative
transmission the videos and images to huge of data and study has been done on the video compression techniques
numerous information. [1] which shows the advance and improvement of video coding
efficiency. More recent studies have been presented and
II. LITERATURE REVIEW discussed in details which focus on the format, properties and
performance of different techniques of images and videos
The research of image and video transform as well as compression [4], [5].
compression techniques has been started early since the
beginning of digital image processing, where uncountable A. Types of Compression
research theories, modules, functions and systems have been
designed, developed and suggested, which made this field most Image compression is categorised into lossy and lossless
attractive and active research area through numerous techniques. In the lossless term of image compression, after the
researchers and scientist, hence the majority of the recent compression the image reconstructed is consider numerically
approaches are involves in one of the three classes which are, identical in compare to the original image. But the lossless
the transform coding as first category, the predictive coding as process compression is able to gain a small amount of image
second category, and the vector quantization as the third compression. However, the lossless coding insures that for the
category alternatively where these categories have been decompressed image is exactly corresponding to the original
explained and further discussed through many researchers [1]. image before the process of compression. However, Lossless
Where in 1992 Ronald-.A.Devore have suggested a special method capable to be applied for the aim of compression of
theory for the purpose of studying the image compression another data form, where the loss of data is not passable such as
technique in term of analysis which mainly relies on program executable, and the text document. The lossless
compression of the wavelet decompositions. Where this compression can be use in various application such as medical,
educational purposes [6]. However some examples of the
lossless term compression are the PNG and the GIF [7].

Where the lossy term of image compression aimed to


reduce the quality of the image. However, during this process
an amount of data in comparing to the original data is usually
lost. The applications of the lossy image compression try to
remove the un-necessary or the redundant data. The
reconstructed image from the lossy compression has
degradation which related to the original one. The data image
compression using lossy process is more suitable for many Figure1: Uncompressed Lena image
applications worldwide especially for fast transfer among the
network. The most popular example of lossy term compression is A. Measuring Quality Image
the JPEG. Whereas, this paper focus on comparing JPEG2000 and
SPIHT techniques for image compression. The parameters used to measure for the comparison of
image quality are the mean square error (MSE), signal to noise
ratio (SNR) and peak signal to noise ratio (PSNR).
Joint Photographic Expert Group2000 A method for the estimation of image quality is needed in
order to give a view about how lossy compression methods
JPEG2000 aims to provide features not possible with JPEG modify image quality. We may treat an image as a matrix whose
for a wide range of applications including internet, colour fax, elements are image pixels.
digital photography, remote sensing, mobile, medical imagery, The estimation process is then based on the calculation of
e-commerce. Basic requirements of this method includes distances between appropriate elements of input and output
efficient coding, spatial and quality scalability, complexity, line matrices. In this way, not only comparison of quality of
and block applications and robustness. JPEG 2000 is a wavelet different compression methods is enabled, but also comparison
transform based compression method and it provides various of the results of the same method using different compression
benefits over the JPEG which uses Discrete Cosine Transform ratios.
(DCT). The wavelet compression transforms an image into Denoting that matrix A at the input of the compression
wavelet series that can be stored more efficiently than blocks of system with elements aij, with i {1...M}, j {1...N}, where M
pixels. [8] is the number of image elements in the vertical and N is the
number of image elements in horizontal direction [10]. M x N
Set Partitioning In Hierarchical Tree (SPIHT) is the total number of image elements.
The output of the compression system is the matrix A' with
The SPIHT is an algorithm developed and presented by elements a'ij. The distance between the elements of matrices A
Pearlman to produce best still image coding. SPIHT and A' represents the error or the loss of image quality. Usually,
compression method is the improvement of Embedded Zero the error is larger for higher compression ratios. A user can set
Tree Wavelet (EZW). It uses an effective spatial orientation tree the compression ratio according to the desired image quality,
structure and bit-plane coding method, and can achieve and hence directly influence the data size of the compression
embedded code stream of high compression efficiency, and it image [10].
has high signal to noise ratio and better image restoration
quality. While comparing with EZW, SPIHT algorithm The total reconstruction error is defined as
constructs two different types of spatial zero tree, taken better
advantage of the attenuation law of wavelet coefficients 1 1
2
amplitude. The drawback is adopting the more complex set = (1)
partitioning strategy to reduce the coding rate and store the =0 =0
zero-tree structure of image wavelet coefficients and important
factor information by using three set list and reduce memory The distance between matrices A and A' is frequently
space utilization. [9] calculated using the Mean Square Error:

III. METHODOLGY 1 1
1 2
In this work the compression techniques has been = = (2)

performed in Vcdemo software on uncompressed image (Lena =0 =0
image: 256x256; grayscale; color depth 24 bit) shown in
Figure1. Where M x N is the total number of image pixels, and the
sum is applied to all image elements.
The amplitudes of image elements are in the range [0,2n-1], TABLE I
MSE, SNR and PSNR values for JPEG2000 compression method
where n is the number of bits needed for binary representation
of amplitude of each element in the original image. MSE does Bit rate 0.1 0.2 0.3 0.4 0.5 0.75 1 1.5 3
not consider amplitudes of image elements (it only considers
MSE 216.7 112.9 73 52.8 37.4 20.1 11.7 5.5 1
differences between amplitudes) and it is the reason for
SNR (dB) 11 13.8 15.7 17.1 18.6 21.3 23.7 27 34.3
introducing the Peak Signal to Noise Ratio:
PSNR(dB) 24.8 27.6 29.5 30.9 32.4 35.1 37.5 40.7 48.1

12 1
= 10. 10 ( ) = 20. 10 ( ) (3) TABLE II
MSE, SNR and PSNR values for SPIHT compression method

Bit rate 0.1 0.2 0.3 0.4 0.5 0.75 1 1.5 3


The variable MAXI is the maximum amplitude value of MSE 191.1 103.1 67.4 46.3 34.6 18.3 11 5.2 0.9
image element (pixel). When the amplitude of the image pixel SNR (dB) 11.6 14.2 16.1 17.7 19 21.7 23.9 27.2 35.1
is represented by B bits, MAXI is 2B-1. PSNR(dB) 25.3 28 29.8 31.5 32.7 35.5 37.7 41 48.8
With n=8 bits/image element we can define:

2552 The values for MSE, SNR and PSNR are measured for
= 10. 10 ( ) (4) Lena image along with JPEG2000 and SPIHT compression
method. The values obtained for MSE, SNR and PSNR for the
uncompressed image for the different compression methods
III. EXPERIMENTAL ANALYSIS IMAGE QUALITY USING used are given in Tables I, II.
JPEG2000 AND SPIHT COMPRESSION METHODS
It can be observed from results obtained that as the bit rates
In order to find the quality of the reconstructed images the
are increased the values of MSE decreases for all the
compression is performed using VCdemo compression
compression methods. For lower Bits per Pixel (bpp) MSE
modules. Nine different bit-rates are used: 0.1, 0.2, 0.3, 0.4, 0.5,
values are gradually higher and vice versa. The differences in
0.75, 1, 1.5, and 3 bpp as shown in Figure 2, 3. For Lean image
MSE values of the compressed image is slightly higher for
the difference between the original and reconstructed images
JPEG 2000 when compared with the values obtained for SPIHT
are calculated with Mean Square Error (MSE), Signal to Noise
compression method.
Ratio (SNR) and Peak Signal to Noise Ratio (PSNR).

250
Spiht
JPEG2000

200

150
MSE

Fig 2. JPEG2000 Image Compression method 100

50

0
0 0.5 1 1.5 2 2.5 3
bpp

Fig 4. MSE values obtained by JPEG200 and SPIHT


compression methods
Fig 3. SPIHT Image Compression method The MSE values acquired by Lena image using different
compression methods is show in Figure 4, while analysing
Figure 4 the JPEG2000 compression method yields slightly
higher MSE values compared to SPIHT compression method.
IV. CONCLUSION
40
Spiht In this paper the compression performed by method of
JPEG2000
SPIHT and JPEG2000 are shown the same bit rates from 0.1 to
35
0.8 and for higher bit rates from 1.0 to 3 the SPIHT produces
higher values of MSE, SNR, and PSNR .Generally speaking the
30
SPIHT compression method provides better results than the
JPEG2000 methods in terms of the quality parameters. High
SNR(dB)

25 PSNR values show that the quality of the resulting image is of


higher quality for the tested image.
20

15 ACKNOWLEDGMENT
I would like to express my appreciation and thanks to Dr.
10
0 0.5 1 1.5 2 2.5 3 Nurulfajar bin Abd. Manap for his information and assistance
bpp
though out the course of advance digital signal processing. I
Fig 5. SNR values obtained by JPEG200 and SPIHT would like to appreciate each effort done in Universiti Teknikal
compression methods Malaysia Melaka as well.
REFERENCES
Figure 5 shows the SNR values while compressing are [1] Thomas Sikora, Trends and Perspectives in Image and Video Coding,
Proceedings of the IEEE, Vol. 93, No. 1, 2005.
higher for higher bit rates and vice versa, according to Figure 5
SPIHT and JPEG 2000 produces just about related values and [2] T. V Janahiraman and K. Win, SOM Based Segmentation Method to
there is slight change SNR is higher at SPIHT. Identify Water Region in LANDSAT Images, vol. 2, no. 1, pp. 1318,
2011.
Original image
50
[3] N. P. PratikChavada and KanuPatel, Region of Interest Based Medical
Spiht Image Compression, pp. 112, 2014.
JPEG2000
45 [4] O. S. Tomasz Grajek, Krzysztof Klimaszewski, 20 Years of Progress in
Video Compression from MPEG-1 to MPEG-H HEVC. General View
on the Path of Video Coding Development, 2016.
40 .
[5] M. Abomhara, video compression techniques: an overview. 2010.
PSNR(dB)

35 [6] R. Gonzalez and R. Woods, Digital image processing. 2010..

[7] T. International and J. Of, The International Journal Of Science &


30
Technoledge Various Image Compression Methods: A Survey Abstract
vol. 2, no. 13, pp. 164168, 2014.
25
[8] S. M. Jog and S. D. Lokhande, Embedded Zero-Tree Wavelet ( EZW )
Image CODEC, Int. Conf. Adv. Comput. Commun. Control, no. 44, pp.
20 379383, 2009.
0 0.5 1 1.5 2 2.5 3
bpp
[9] Sanjith, S.; Ganesan, R.; Isaac, R. S. Rimal Experimental Analysis of
Fig 6. PSNR values obtained by JPEG200 and SPIHT Compacted Satellite Image Quality Using Different Compression
compression methods Methods Advanced Science, Engineering and Medicine, Volume
7, Number 3, March 2015, pp. 227-233(7)

In Figure 6 The PSNR values increases while the bitrates


are increased image SPIHT and JPEG 2000 has nearly similar
values .However, there is a slightly change in both compression
methods when comparing both methods it can be observed
improvement by using Spiht compression techniques over
JPEG2000 method.

S-ar putea să vă placă și