VOL. 13, NO. 5, MARCH 2018 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
©2006-2018 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1877
MEDICAL IMAGE DATA COMPRESSION USING HYBRID METHODS
Alyaa H. Ali
Department of Physics, College of Science, University of Baghdad, Baghdad, Iraq
E-Mail: aliahusain@ymail.com
ABSTRACT
This search focuses on Image Data compression using different methods for data compression, three images are
simulated for this technique based on modified method, Huffman with local and soft threshed and three block size 4×4,
8×8 and 16×16. The second one is based on using the DCT "discrete cosine transformation" and "discrete wavelet
transformation" DWT these methods are applied on the stroke brain images" Computing tomography CT images " after
using the region based segmentation method to get the region of interest which is the stroke, Completing the process by
calculating quality of image compression, five parameter are used ," Peak Signal to Noise Ratio (PSNR)", "Mean Square
Error (MSE)","Compression Ratio(CR)","Structural Similarity Index (SSIM)" and "Universeral Image quality Index
(UIQI)".
Keyword: DCT, DWT, Huffman coding, region based method.
INTRODUCTION
In medical image processing the need for the
image compression is essentially important and can
divided into two structure, the first one is the lossy image
compression higher value of CR "compression Ratio is
obtained but, the compressed image is Distorted
comparing it to the original image while the lossless image
compression is identical with the original one [1]. The
need for the image compression require considerable
storage capacity, since the medical images such as
Computed Tomography CT, X-ray and Magnetic there
typical size range are large and the challenge to limited
band compression is one of the methods used to reduce
the image size. Different methods are developed all these
method tend to achieving high quality of decompressed
image, the compression can be accomplished either by
lossy or lossless methods [2].
The image compression technique is used to save
the important data which used only in the analysis. The
image compression can be divided into two phrase, the
first one is known as the lossless, in this one no
information is lost, the original image is completely the
same as the compressed image. The most known one is the
Huffman coding. The lossy one, the original image cannot
match the compressed image, this compression techniques
reach image compression by losing part of the information
while keeping the reconstruction quality. Hence, the data
cannot be restore exactly the same as the original image
[3]. The most commune methods is the Discrete Cosine
Transformation and Wavelet Transformation [4] the image
compression decrease the size of image storage bite
without losing the quality of the image, also the time
required to store or send the image decrease. The edge and
the pixels which repeated be decrease by image
compression [5]. The DCT based on separating the
compressed image into different frequency parts, it
consider to be lossy because the first part of the
compression called quantization and the other which is
not important frequency are removed [6] so, the "DCT"
basics idea depend on "disintegrating the image into
segment" [7], the lossy data compression give best
compression ratio[8]. The wavelet is early appear in 1990,
its idea is based on dividing the information of the image
into four sub-bands in which the image is transformed into
"low image information" and the remaining details are in
three images " Horizontal, Vertical, and Diagonal" images.
The first one is also decomposed into other four sub-
images, this process gives a number of features which
cannot be seen in the original images and can appear in the
level after the transformation so, for this the "wavelet
transformation" is the best for medical image compression.
The Huffman coding which is produced by
Doctor David. A. Huffman in 1952 it is a tool to build the
image with less redundancy coding, the Huffman coding is
a statistical coding, it try to minimize the number of the
bits which is needed to represent the symbols [3]. These
method is applied in this search the region of interest
which is obtained using the "Base Region Method" which
is the selection of the region of interest to compressed in
the medical images of the diagnostic region, this method is
used to have balance between the very good quality of the
reconstructed and low memory Douckas et al. in 2007 has
used the "Volumetric medical images" to analysis different
region of interest [9].
Test images: The images which are used in the
search are three brain stroke Gray Images of size 255×255,
JPG type, (a) shows the gray scale image without skull and
some noise in the image, (b) shows the image after using
morphological properties "Erosion and Dilation " to
remove the outer skull, (c) gives the region of interest
which is the stroke after using the "region based
method"[10].