IMAGE COMPRESSION BY EMBEDDING FIVE MODULUS METHOD INTO JPEG

Signal & Image Processing : An International Journal (SIPIJ) Vol.4, No.2, April 2013
DOI : 10.5121/sipij.2013.4203 31
IMAGE COMPRESSION BY EMBEDDING FIVE
MODULUS METHOD INTO JPEG
Firas A. Jassim
Management Information Systems Department,
Irbid National University, Irbid 2600, Jordan
Firasajil@yahoo.com
ABSTRACT
The standard JPEG format is almost the optimum format in image compression. The compression ratio in
JPEG sometimes reaches 30:1. The compression ratio of JPEG could be increased by embedding the Five
Modulus Method (FMM) into the JPEG algorithm. The novel algorithm gives twice the time as the standard
JPEG algorithm or more. The novel algorithm was called FJPEG (Five-JPEG). The quality of the
reconstructed image after compression is approximately approaches the JPEG. Standard test images have
been used to support and implement the suggested idea in this paper and the error metrics have been
computed and compared with JPEG.
KEYWORDS
Image compression, JPEG, DCT, FMM, FJPEG.
1. INTRODUCTION
The main goal of image compression methods is to represent the original images with fewer bits.
Recently, image compression is very popular in many research areas. According to the research
area, one of the two types of compression, which are lossless and lossy, can be used [13].
Lossless compression can retrieve the original image after reconstruction. Since it is impossible to
compress the image with high compression ratio without errors, therefore; lossy image
compression was used to obtain high compression ratios. Consequently, reducing image size with
lossy image compression gives much more convenient ratio than lossless image compression [7]
[9].
Since the mid of the eighties of the last century, the International Telecommunication Union
(ITU) and the International Organization for Standardization (ISO) have been working together to
obtain a standard compression image extension for still images. The recommendation ISO DIS
10918-1 known as JPEG Joint Photographic Experts Group [11]. Digital Compression and
Coding of Continuous-tone Still Images and also ITU-T Recommendation T.81 [11]. After
comparing many coding schemes for image compression, the JPEG members selected a Discrete
Cosine Transform (DCT). JPEG became a Draft International Standard (DIS) in 1991 and an
International Standard (IS) in 1992 [12]. JPEG has become an international standard for lossy
compression of digital image.

32
2. FIVE MODULUS METHOD
Five modulus method (FMM) was first introduced by [6]. The main concept of this method is to
convert the value of each pixel into multiples of five. This conversion omits parts of the signal
that will not be noticed by the signal receiver namely the Human Visual System (HVS). Since the
neighbouring pixels are correlated in image matrix, therefore; finding less correlated
representation of image is one of the most important tasks. The main principle of image
compression states that the neighbours of a pixel tend to have the same immediate neighbours [4].
Hence, the FMM technique tends to divide image into 8×8 blocks. After that, each pixel in every
block can be transformed into a number divisible by 5. The effectiveness of this transformation
will not be noticed by the Human Visual System (HVS) [6]. Therefore, each pixel value is from
the multiples of 5 only, i.e. 0, 5, 10, 15, 20, … , 255. The FMM algorithm could be stated as:
if Pixel value Mod 5 = 4
Pixel value=Pixel value+1
Pixel value=Pixel value+2
Pixel value=Pixel value-2
Pixel value=Pixel value-1
Now, to illustrate the method of Five Modulus Method (FMM). An arbitrary 8×8 block has been
taken randomly from an arbitrary digital image and showed in table 1.
Table 1. Original 8×8 block
106 98 104 102 109 110 107 113
103 107 104 110 109 110 110 113
106 106 105 110 111 107 104 108
104 105 110 111 109 108 110 104
106 106 119 113 111 107 109 108
106 104 101 105 104 104 107 113
97 103 104 101 102 104 106 110
103 106 110 105 103 105 103 108
After that, the FMM algorithm shown earlier may be applied to the 8×8 block in table (1). Every
pixel value was converted into multiple of five, i.e. the first pixel which is (106) may be
converted into (105), etc. Therefore, the new resulting 8×8 block was showed in table (2).
Table 2. Converting 8×8 block by five modulus method (FMM)
105 100 105 100 110 110 105 115
105 105 105 110 110 110 110 115
105 105 105 110 110 105 105 110
105 105 110 110 110 110 110 105
105 105 120 115 110 105 110 110
105 105 100 105 105 105 105 115
95 105 105 100 100 105 105 110
105 105 110 105 105 105 105 110

33
Now, to complete the FFM method, the 8×8 block shown in table (2) may be divided by 5 to
reduce the pixel values into a lesser values. Therefore, the first converted pixel (105) would
be (105/5=21), etc. The evaluated 8×8 block after division is shown in table (3).
Table 3. Dividing 8×8 block in table (1) by 5
21 20 21 20 22 22 21 23
21 21 21 22 22 22 22 23
21 21 21 22 22 21 21 22
21 21 22 22 22 22 22 21
21 21 24 23 22 21 22 22
21 21 20 21 21 21 21 23
19 21 21 20 20 21 21 22
21 21 22 21 21 21 21 22
The main concept of the FMM method is to reduce the dispersion (variation) between pixel
values in the same 8×8 block. Hence, the standard deviation in the original 8×8 block was
(3.84) while it was (0.85) in the transformed 8×8 block. This implies that, the storage space
for the transformed 8×8 block will be less than that of the original 8×8 block.
3. FJPEG ENCODING AND DECODING
The basic technique in FJPEG encoding is to apply FMM first. Each 8×8 block is transformed
into multiples of 5, i.e. Five Modulus Method (FMM), see table (2). After that, dividing the
whole 8×8 block by 5 to obtain new pixel values range [0..51] which are the results of
dividing [0..255] by 5. Now, the 8×8 block is ready to implement the standard JPEG
exactly starting with DCT and so on. The new file format if FJPEG which is the same as
JPEG but all its values are multiples of 5. Actually, the FJPEG image format could be sent
over the internet or stored in the storage media, or what ever else, with a lesser file size as a
compressed file. On the other hand, at the decoding side, when reconstructing the image,
firstly, the IFMM (Inverse Five Modulus Method) could be applied by multiplying the 8×8
block by 5 to retrieve, approximately, the original 8×8 block. After that, the same JPEG
decoding may be applied as it is. Therefore, the main contribution is this article is to embed
FMM method into JPEG to reduce the file size. The whole encoding and decoding
procedures may be shown in figure (1).
One of the main advantages in FJPEG compared to JPEG is that the reduction in file size is
noticeable. Unfortunately, one of the main disadvantages is that on the decoding side and before
applying JPEG decoding the image must be multiplied by 5. Theses calculations will be evaluated
on the computer processor at the decoding side. Actually, these calculations will surely offtake
some time and space from the memory and CPU.

34
Figure 1. Encoding and Decoding procedures in FJPEG
4. DISCRETE COSINE TRANSFORM
Discrete Cosine Transformation (DCT), is used to compress digital images by reducing the
number of pixels used to express 8x8 blocks into a lesser number of pixels. Nowadays, JPEG
standard uses the DCT as its essentials. This type of lossy encoding has become the most popular
transform for image compression especially in JPEG image format. The origin of the DCT back
to 1974 by [1]. The DCT algorithm is completely invertible which makes it useful for both
lossless and lossy image compression. The DCT transform the pixel values of the 8×8 block into
two types. The first type is the highest values that can be positioned at the upper left corner of the
8×8 block. While the second type represents the values with the smaller value which can be
found at the remaining areas of the 8×8 block. The coefficients of the DCT can be used to
reconstruct the original 8×8 block through the Inverse Discrete Cosine Transform (IDCT) which
can be used to retrieve the image from its original representation [7].
Images can be separated by DCT into segments of frequencies where less important frequencies
are omitted through quantization method and the important frequencies are used to reconstruct the
image through decoding technique [8]. The main two advantages for embedding FMM into
JPEG are:
• Reducing the dispersion (variation) degree between DCT coefficients.
• Decreasing the number of the non-zero elements of the DCT coefficients.
According to [3], the measure DCT-STD (STD is the standard deviation) have been used to
measure the clarity of am image. Here, the DCT-STD have been used to measure the difference
of the DCT of the 8x8 block taken from the traditional JPEG image with 8x8 block taken from
FJPEG image, tables 4 and 5 respectively.

35
Table 4. DCT transform of the original block
853 -10 -2 -6 2 0 2 0
7 -3 -2 6 4 0 0 0
-8 -5 6 1 0 -4 3 -1
0 -5 5 0 1 0 2 2
4 4 -4 -1 -3 3 5 5
-8 -3 1 3 0 0 0 1
-1 2 3 3 5 -1 1 0
1 2 -3 -2 0 -1 3 3
Table 5. DCT transform of the FMM block
170 -2 0 -1 0 0 0 0
1 0 0 1 1 0 0 0
-1 -1 1 0 0 -1 0 0
0 0 0 0 0 0 0 0
1 0 0 0 0 0 0 1
-1 0 0 0 0 0 0 0
0 0 0 0 1 0 0 0
0 0 0 -1 0 0 0 0
The value of the standard deviation for table (4) is 106.65, while the standard deviation for table
(5) is 21.26. Actually, when the standard deviation value is high, it means that the difference
between most values and their mean values are high; and vice versa. Hence, coefficients of table
(4) have high variation than table (5). It means that, the dispersion obtained by the transformed
FMM is less than the traditional dispersion of the original DCT. Also, it can be seen that the
number of non-zero elements of table (2) are lesser than of those of table (1). According to [5],
every zero element of the DCT matrix saves operations and reduces the time complexity. Also,
the total processing time required for IDCT is decreased when the number of the non-zero
elements of the DCT coefficient matrix is less [10]. Actually, in image compression, few non-
zero elements of the DCT coefficients could be used to represent an image [2].
Fortunately, the previous mentioned opinions decants in the behalf of the proposed idea in
this article which is that the reduction of the non-zero elements of the DCT matrix will
reduce the time complexity. Hence, using DCT obtained by the FMM method will gives
sophisticated results than using the traditional DCT that contains more non-zero elements.
5. EXPERIMENTAL RESULTS
Actually, the proposed FJPEG in this article has been implemented to a variety of standard test
images with different spatial and frequency characteristics. As mentioned earlier, the embedding
of FMM into JPEG will formulate FJPEG which has a noticeable difference in image size.
Experimentally, the compression ratio between JPEG and FJPEG has been showed in table (6).
Also, PSNR between JPEG and FJPEG was calculated in the same table.

36
Table 6. File size, CR and PSNR for test images
File Size (KB) CR PSNR
JPEG FJPEG JPEG FJPEG JPEG FJPEG
Lena 36.9 13.5 20.8:1 56.9:1 37.4882 32.3178
Baboon 75.6 27.6 10.2:1 28.8:1 30.7430 25.0817
Peppers 40.5 14.3 19.0:1 53.7:1 35.5778 31.6617
F16 26 15 29.5:1 51.2:1 Inf Inf
Bird 131 27 7.7:1 37.5:1 53.4068 30.4266
ZigZag 64.2 23.5 5.6:1 15.2:1 31.0531 20.1237
Houses 66.8 24.2 8.9:1 24.5:1 Inf 21.9970
Mosque 23.9 8.33 14.7:1 42.1:1 36.1242 28.6154
Figure 2. Variety of test images
As an example, three test images (Lena, baboon, and Peppers) have been used, figures 3,4, and 5,
to show the visual and storage differences between (a) standard JPEG (b) FJPEG after encoding
directly, i.e. this file will be sent over the internet and stored in the storage media which has a
small file size compared to JPEG (c) FJPEG after reconstruction, i.e. multiplied by 5 on the
decoding side.
As seen from figures 3,4, and 5, that the differences by the human eye are worthless and can not
be recognized visually. As a quantitative measure, the PSNR has been used to compute the signal-
to-noise-ratio for the original JPEG and the FJPEG and the result are presented in table (6). The
differences between PSNR for the two images are reasonable. As an example, for Lena image the
PSNR of the JPEG was (37.4882) while the PSNR for the FJPEG was (32.3178) which
is not very high difference. Therefore, one could accept this fair difference by taking
into consideration the high difference in file size (36.9 KB) for JPEG while (13.5
KB) for the FJPEG.

37
(a) (b) (c)
Figure 3. (a) JPEG (36.9 KB) (b) FJPEG (sent) (13.5 KB) (c) FJPEG×5 (reconstructed) (24 KB)
(a) (b) (c)
Figure 4. (a) JPEG (75.6 KB) (b) FJPEG (sent) (26.7 KB) (c) FJPEG×5 (reconstructed) (50.6 KB)
(a) (b) (c)
Figure 5. (a) JPEG (40.5 KB) (b) FJPEG (sent) (14.3 KB) (c) FJPEG×5 (reconstructed) (25.9 KB)
The main difference in compression ratio between FJPEG and JPEG may be illustrated
graphically in figure (6).

38
0
10
20
30
40
50
60
Lena
Baboon
Peppers
F16
Bird
ZigZag
Houses
Mosque
CR(x:1)
JPEG
FJPEG
Figure 6. Difference in compression ratio between JPEG and FJPEG
6. CONCLUSIONS
The method of FJPEG stated in this article can be used as a redundancy alternative of the
traditional JPEG in most of the disciplines of science and engineering especially in image
compression because of its noticeable difference in image size or the compression ratio. As a first
recommendation, the research area of the CPU time and space taken to reconstruct the FJPEG at
the encoding side needs a lot of interest. As a second recommendation, the implementation of
FJPEG in video compression needs to be discussed and researched. As a third recommendation,
the alternative name of JPEG method in audio is called MP3, therefore; using the FJPEG in audio
compression could be researched.
REFERENCES
[1] Ahmed, N., T. Natarajan, and R. K. Rao (1974) “Discrete Cosine Transform,” IEEE Transactions on
Computers, C-23:90–93.
[2] Aruna Bhadu, Vijay Kumar, Hardayal Singh Shekhawat, and Rajbala Tokas, An Improved Method of
feature extraction technique for Facial Expression Recognition using Adaboost Neural Network,
International Journal of Electronics and Computer Science Engineering (IJECSE) Volume 1, Number
3 , pp: 1112-1118, 2012.
[3] Chu-Hui Lee and Tsung-Ping Huang, Comparison of Two Auto Focus Measurements DCT-STD and
DWT-STD, Proceeding of the International MultiConference of Engineers and Computer Scientists,
Vol I, IMECS, March 14, 2012, Hong Kong.
[4] David Salomon. Data Compression: The Complete Reference. fourth edition, Springer 2007.
[5] Deepak Nayak, Dipan Mehta, and Uday Desai, A Novel Algorithm for Reducing Computational
Complexity of MC-DCT in Frequency-Domain Video Transcoders, ISCAS In International
Symposium on Circuits and Systems (ISCAS 2005), 23-26 May 2005, Kobe, Japan. pages 900-903,
IEEE, 2005.
[6] Firas A. Jassim and Hind E. Qassim, Five Modulus Method For Image Compression, Signal & Image
Processing : An International Journal (SIPIJ) Vol.3, No.5, October 2012
[7] Gonzalez R. C. and Woods R. E., “Digital Image Processing”, Second edition, 2004.
[8] Ken Cabeen and Peter Gent, "Image Compression and the Discrete Cosine Transform” Math
45,College of the Redwoods.
[9] Khalid Sayood, Introduction to Data Compression, third edition. Morgan Kaufmann, 2006.

39
[10] Lee J. and Vijaykrishnan N., and M. J. Irwin, Inverse Discrete Cosine Transform Architecture
Exploiting Sparseness And Symmetry Properties, Embedded & Mobile Computing Design Center
The Pennsylvania State University, NSF ITR 0082064.
[11] Tinku Acharya and Ping-Sing Tsai, JPEG2000 Standard for Image Compression Concepts,
Algorithms and VLSI Architectures, JOHN WILEY & SONS, 2005.
[12] Wallace G. K., The JPEG Still Picture Compression Standard, Communication of the ACM, Vol. 34,
No. 4, 1991, pp. 30-44.
[13] Zhang H., Zhang ., and Cao S., “Analysis and Evaluation of Some Image Compression Techniques,”
High Performance Computing in Asia- Pacific Region, 2000 Proceedings, 4th Int. Conference, vol. 2,
pp: 799-803, May, 2000.
Authors
Firas Ajil Jassim was born in Baghdad, Iraq, in 1974. He received the B.S. and M.S.
degrees in Applied Mathematics and Computer Applications from Al-Nahrain
University, Baghdad, Iraq, in 1997 and 1999, respectively, and the Ph.D. degree in
Computer Information Systems (CIS) from the Arab Academy for Banking and
Financial Sciences, Amman, Jordan, in 2012. In 2012, he joined the faculty of the
Department of Business Administration, college of Management Information System,
Irbid National University, Irbid, Jordan, where he is currently an assistance professor.
His current research interests are image compression, image interpolation, image
segmentation, image enhancement, and simulation.

IMAGE COMPRESSION BY EMBEDDING FIVE MODULUS METHOD INTO JPEG

More Related Content

What's hot (19)

Viewers also liked (19)

Similar to IMAGE COMPRESSION BY EMBEDDING FIVE MODULUS METHOD INTO JPEG (20)

Recently uploaded (20)

IMAGE COMPRESSION BY EMBEDDING FIVE MODULUS METHOD INTO JPEG