AN OPTIMIZED BLOCK ESTIMATION BASED IMAGE COMPRESSION AND DECOMPRESSION ALGORITHM

http://www.iaeme.com/IJCET/index.asp 9 editor@iaeme.com
International Journal of Computer Engineering & Technology (IJCET)
Volume 7, Issue 1, Jan-Feb 2016, pp. 09-17, Article ID: IJCET_07_01_002
Available online at
http://www.iaeme.com/IJCET/issues.asp?JType=IJCET&VType=7&IType=1
Journal Impact Factor (2016): 9.3590 (Calculated by GISI) www.jifactor.com
ISSN Print: 0976-6367 and ISSN Online: 0976–6375
© IAEME Publication
___________________________________________________________________________
AN OPTIMIZED BLOCK ESTIMATION
BASED IMAGE COMPRESSION AND
DECOMPRESSION ALGORITHM
JNVR Swarup Kumar
Assistant Professor, Department of CSE,
Gudlavalleru Engineering College,
Gudlavalleru, India
R Deepika
PG Student, Department of CSE,
Gudlavalleru engineering college, Gudlavalleru, India
ABSTRACT
In this paper, we implemented a new model of image compression and
decompression method to search the aimed image based on the robust image
block variance estimation. Many methods of image compression have been
proposed in the literature to minimize the error rate and compression ratio.
For encoding the medium type of images, traditional models use hierarchical
scheme that enables the use of upper, left, and lower pixels for the pixel
prediction, whereas the conventional raster scan prediction methods use
upper and left pixels. In this proposed work, we have implemented block
estimation and image distortion rate to optimize the compression ration and to
minimize the error rate. Experimental results show that proposed model gives
a high compression rate and less rate compared to traditional models.
Key words: Compression, Block Size, Probability, Encode, Decode.
Cite this Article: JNVR Swarup Kumar and R Deepika. An Optimized Block
Estimation Based Image Compression and Decompression Algorithm.
International Journal of Computer Engineering and Technology, 7(1), 2016,
pp. 09-17.
http://www.iaeme.com/IJCET/issues.asp?JType=IJCET&VType=7&IType=1
1. INTRODUCTION
Image compression can be defined [1] in a simple manner as an application of data
compression that minimizes the size of the original image without resulting in the
degradation of image quality. The principal approach in data compression is the
reduction in the amount of image data (the number of bytes) while preserving

JNVR Swarup Kumar and R Deepika
information (image details). Hence, image compression aims to reduce both the
irrelevance and the redundancy available in the image data with the intention of
optimizing and putting to maximum use the data storage and data transmission
facilities. Here, the irrelevancy reduction implies that the information removed in this
process is systematically selected such that it includes data irrelevant to the user. Such
reduction strategy, where the emphasis is on the „meaning‟ of the information, leads
to lossy compression. Redundancy reduction, on the other hand, may be employed for
lossless compression as it is based on data statistics and leads to the reduction of the
reiteration of the same bit patterns in the data. Since image compression addresses the
problems of the rate of data transfer (Bandwidth requirements) as well as data storage
(Space requirements), its applications and extensive usage are seen in varied fields
such as law enforcement, internet applications, medicine, weather predictions, earth
resources tracking and management, satellite imagery and countless more. So, in spite
of the fast-paced developments taking place regarding superior processing capabilities
in terms of the speed of processing, the volume of data storage available and the
system performance in general, the ever-expanding needs of the digital media are
consistently contributing to the shortage of these facilities. The long strides taken
forward due to technological advances still leave space for more development. The
demand for better performance is persistent. This situation attests the need for better
and better image compression techniques. Hence, there is a need to develop image
compression algorithms which ensure improved performance. The performance of
image compression algorithms can be measured in terms of metrics such as CR, MSE,
PSNR, etc. The improvement in these parameters must also be accompanied by the
basic requirement of any application employing compression procedures, i.e., No loss
of relevant information and No degeneration in image quality.
The inherent quality of the wavelets can be exploited to meet the diverse and
heterogeneous requirements regarding the bandwidth and computational capabilities
of individual internet users. In this case, the low end users are provided coarse
approximations while the high end users are allowed to utilize maximum bandwidth
to meet their fidelity criteria[1-5].
A set of variable-size codes to symbols based on their probabilities. Huffman
coding produces ideal variable-size codes (codes whose average size equals the
entropy) when the symbols have probabilities of occurrence that are negative powers
of 2. This is because the Huffman method assigns a code with an integral number of
bits to each symbol in the alphabet.
Raw image and video data bitrate requirements

An Optimized Block Estimation Based Image Compression and Decompression Algorithm
Huffman coding is widely used, but is inefficient when the alphabet size is small or
symbol probabilities are highly skewed. AVS standard uses a context based lossless
coding scheme, namely C-2D-VLC which uses multiple two-dimensional variable
length coding tables along with Exponential- Golomb codes to encode symbols. The
two-dimensional tables are made adaptive based on contexts [5-7].
2. LITERATURE SURVEY
Lossy image compression techniques can be broadly classified into two categories,
namely spatial domain techniques and transform domain techniques. In spatial
domain techniques, the pixels in the image are used, whereas in transform domain
techniques the image pixels are converted into a new set of values, namely transform
coefficients, for further processing. Predictive coding is a well known spatial domain
technique that operates directly on the image pixels. It consists of three main
components, namely quantizer, predictor and entropy coder. The drawback of this
technique is that it causes more blurring of image edges and distortions in smoother
areas. Also, the design process is complex in the sense that the predictor must be
designed assuming zero quantization error and quantizer must be designed to reduce
its own error. Transform coding is a widely used method in lossy image compression.
An image is compressed by transforming the correlated pixels to a new representation
(transform domain) where they are decorrelated, that is, the transform coefficients are
independent of one another and most of the energy is packed in a few coefficients.
The transform coefficients are quantized to reduce the precision of these values so
that very few quantized nonzero coefficients have to be encoded. This is a many-to-
one mapping, and consequently a lossy process, which provides the main source of
compression. Quantized coefficients are further compressed using entropy coding
techniques to provide a better overall compression. The transform used is a linear
transform so that the task of compressing in the transform domain is more efficient
and simple. There are many hybrid image compression techniques that are based on
DWT and JPEG standard The wavelet coefficients in each subband are JPEG coded in
different ways. In [4] the input image is first transformed by Discrete Cosine
Transform (DCT) and the lowest frequency components are split into sixteen fine
uniform subbands by considering correlation and energy in the frame. This algorithm
offers high CR and good subjective quality compared to JPEG standard for any
specified value of PSNR. The Log-Exp image compression algorithm operates pixel-
by-pixel without producing any blocking effects as in JPEG standard. This algorithm
uses Huffman coding technique which is a static technique and consumes more time.
Also, Huffman coding is inefficient when the symbol probabilities are highly skewed.
Block Truncation Coding (BTC) techniques are known for their speed and reduced
computational complexity since complicated transforms are not used The principle
used in BTC algorithm is to use two-level quantizer on each non-overlapping 4 BY 4
blocks of the image. The quantizer adapts to local properties of the image while
preserving the first- or first- and second-order statistical moments. The BTC
technique of Delp preserves the first- and second-order moments. In the Modified
BTC (MBTC) of Udpikar (Udpikar and Raina 1987), only the first-order moments are
preserved. The parameters transmitted or stored in the BTC based algorithms are the
statistical moments and the bitplane. The BTC techniques yield good quality images
at a bitrate of 2 bpp or a CR of 4. [8] have proposed “an interpolative coding approach
using JPEG technology with the aim of reducing the visually unpleasant blocking
artifacts produced by JPEG at low bit rates” or higher compression levels. When
compared with JPEG, this scheme has achieved higher PSNR at low bit rates.

Bruckstein, Elad and Kimmel have also worked in the same direction in order to
improve on the objective as well as the subjective performance of JPEG. They have
employed a down sampling approach based on image statistics, size and the available
bit budget, prior to JPEG coding. This approach has several advantages. Since the
down sampling is done prior to coding, the computational complexity in the
coder/decoder is decreased significantly. The visual quality and the PSNR
performance are also enhanced considerably. Also, this approach allows an expansion
in the range of the employable bit-rates in compressing the image satisfactorily.
However, this work is limited in that it uses fixed filters for decimation and
interpolation. Inspired by the above work, Tsaig et al. have improved upon it by
introducing optimal filters for decimation along with interpolation stages, thereby
achieving significant gain in both qualitative and quantitative metrics. Though all
these schemes have good proposals, they are limited in the under stated aspects: a)
The down sampling ratio is fixed by the user; b)the critical bit rate is low and is
imagedependent; c)the encoder has to switch between a down sampling scheme and
traditional scheme to give a good coding quality for different images[8-11].
3. PROPOSED WORK
Traditional Model Image Decomposition
For more accurate prediction of these signals, and also for accurate modeling of the
prediction errors, we use the hierarchical scheme: the chrominance image is
decomposed into two subimages; i.e. a set of even numbered rows and a set of odd
numbered rows respectively. Once the even row subimage Xe is encoded, we can use
all the pixels in Xe for the prediction of a pixel in the odd row subimage Xo. In
addition, since the statistical properties of two subimages are not much different,
the pdf of prediction errors of a subimage can be accurately mod- eled from the
other one, which contributes to better context modeling for arithmetic coding as
shown in Fig.
For the efficient compression, the statistics of symbols (prediction errors)
should well be described by an appropriate model and/or parameters. Traditional
model used the prediction error as a random variable with pdf P(e|Cn ), where Cn is
the coding context that reflects the magnitude of edges and textures. Specifically, Cn
is the level of quantization steps of pixel activity σ(i, j) defined as
σ(i, j) = |xe(i, j) − xe(i + 1, j)|.

Proposed Model Modification
Step 1: Read Color Image.
Step 2: RGB is first transformed to YCuCv.
Step 3: Image RGB components transformation
Step 4: for each block size NXN.
Find the block variance
Step 5: Each block is estimated from the previous frame using Sum of absolute
difference.
1 1
0 0
(min) min | x(i, j) ( ( , ))( 1, 1)|
N N
i j
SAD Cummprob x i j i j
 
 
   
Cummprob =
0.5* ( ( , ) )/x i j median scale 
Step 6:
dev = x - median;
Density= (1 / PI) * (scale / (dev * dev + scale * scale));
Step 7: After the color transformation, the luminance channel Y is compressed by a
JPEG-LS coder. Pixels in chrominance channels are predicted by the traditional
hierarchical decomposition and directional prediction.
Step 8: Calculate the distortion measure, according to the following formula
2
, 0
( )
N
i j
D p p


 
Finally, an appropriate context modeling of prediction residuals is introduced and
arithmetic coding is applied.

4. EXPERIMENTAL RESULTS
All experiments are performed with the configurations Intel(R) Core(TM)2 CPU
2.13GHz, 2 GB RAM, and the operating system platform is Microsoft Windows XP
Professional (SP2).
Figure 1 Home view of Proposed Approach
Figure 2 Input image
Figure 3 Existing Compression Model

Compressed Bit Rate (CBB): 14.97587
Compressed Time: 0.89881
PSNR RATIO: 0.96389
MSE: 0.28049
Decompression
Proposed Model Compression Results
Compressed Bit Rate (CBB): 9.28099
Compressed Time: 0.71977
Image Size
(KB)
Existing Model
Compression Rate (%)
Proposed Model
Compression Rate (%)
15 14.56 8.95
20 15.34 9.76
30 18.25 11.56

Image Size (KB) Existing Time (sec) Proposed Time (sec)
15 12.67 6.78
20 15.34 8.45
30 18.34 10.56
CONCLUSION
Propoed model compress high dimensional images with high compression rate and
less distortion rate. For encoding the medium type of images, traditional models use
hierarchical scheme that enables the use of upper, left, and lower pixels for the pixel
prediction, whereas the conventional raster scan prediction methods use upper and left
pixels. In this proposed work, we have implemented block estimation and image
distortion rate to optimize the compression ration and to minimize the error rate.
Experimental results show that proposed model gives a high compression rate and less
rate compared to traditional models.
REFERENCES
[1] M. Effros, K. Viswesvariah, S. Kulkarni, S. Verdú. Universal lossless source
coding with the Burrows Wheeler transform. IEEE Transactions on Information
Theory, 48(5):10611081, 2002. 615
[2] T. S. Han, K. Kobayashi. Mathematics of Information and Coding. American
Mathematical Society, 2002. 615

[3] L. T. W. Alexander P. Morgan and R. A. Young, A gaussian derivative based
version of jpeg for image compression and decompression, in IEEE Transactions
on Image Processing, vol. 7, no. 9, September, 1998, pp. 1311–1320.
[4] I. Vilovic, An experience in image compression using neural networks, in 48th
International Symposium ELMAR- 2006, Zadar, Crotia, 07-09, June 2006, pp.
95–98.
[5] Jau-Ji Shen and Hsiu-Chuan Huang, an Adaptive Image Compression Method
Based on Vector Quantization, IEEE, pp. 377-381, 2010.
[6] Suresh Yerva, Smita Nair and Krishnan Kutty, Lossless Image Compression
based on Data Folding, IEEE, pp. 999-1004, 2011.
[7] Firas A. Jassim and Hind E. Qassim, Five Modulus Method for Image
Compression, SIPIJ Vol.3, No.5, pp. 19-28, 2012.
[8] Ashok Kumar, Rajiv Kumaran, Sandip Paul and Sanjeev Mehta. A Low Complex
ADPCM Image Compression Technique with Higher Compression Ratio.
International Journal of Computer Engineering and Technology, 4(6), 2013, pp.
367-377.
[9] Mridul Kumar Mathur, Seema Loonker and Dr. Dheeraj Saxena ―Lossless
Huffman Coding Technique For Image Compression And Reconstruction Using
Binary Trees, IJCTA, pp. 76-79, 2012.
[10] V.K Padmaja and Dr. B. Chandrasekhar, Literature Review of Image
Compression Algorithm, IJSER, Volume 3, pp. 1-6, 2012.
[11] Ch. Ramesh, Dr. N.B. Venkateswarlu, Dr. J.V.R. Murthy. A Novel K-Means
Based Jpeg Algorithm For Still Image Compression. International Journal of
Computer Engineering and Technology, 3(1), 2012, pp. 339-354.
[12] Kaur M. and Kaur G, A Survey of Lossless and Lossy Image Compression
Technique, International Journal of Advanced Research in Computer Science and
Software Engineering Volume 3, Issue 2, February 2013
[13] Somasundaram K. and Domnic S. Modified Vector Quantization Method For
Image Compression, Proceeding of World Academy Of Science, Engineering and
Technology Volume 13 May 2006.
[14] S. Anandanarayanan and Dr. S. K. Srivatsa. A High Performance Novel Image
Compression Technique Using Huffman Coding with EDGE Dection.
International Journal of Computer Engineering and Technology, 4(2), 2013, pp.
17-22.

AN OPTIMIZED BLOCK ESTIMATION BASED IMAGE COMPRESSION AND DECOMPRESSION ALGORITHM

Recommended

More Related Content

What's hot (20)

Viewers also liked (14)

Similar to AN OPTIMIZED BLOCK ESTIMATION BASED IMAGE COMPRESSION AND DECOMPRESSION ALGORITHM (20)

More from IAEME Publication (20)

Recently uploaded (20)

AN OPTIMIZED BLOCK ESTIMATION BASED IMAGE COMPRESSION AND DECOMPRESSION ALGORITHM