الفهرس 
Background & Related Work
Environment SettingsOnly 14 pages are availabe for public view
 |  | from | 32 |  |  |
| | | from | 32 | | |
Abstract
Image De-noising is a subfield of image enhancement in general, which it focuses on the
removal of any undesired details that corrupts the digital image. Actually, this can be
achieved through various filtering techniques, where the variation is compared on a base of
enhancement parameters and the keep of sensitive and important details. On the other hand,
medical imaging is the technique used to create images of the human body or parts of it for
clinical purposes. Medical images always have large sizes and they are commonly corrupted
by single or multiple noise type at the same time, due to various reasons, these two reasons
are the triggers for moving toward parallel image processing to find alternatives of image denoising techniques.
This thesis proposes hybrid de-noising approach that is based on adaptive median filter in the
spatial domain followed by wiener filter in the Fourier transform domain for the removal of
circular blurredness, Gaussian and impulse additive noises simultaneously. The proposed denosing approach is tested on a data set of gray scale medical images or Digital Imaging and
Communications in Medicine (DICOM), each of which was corrupted by additive Gaussian
noise with variance 0.05 and Salt & pepper with probability 0.2. Moreover, we analyze the
hybrid de-noising approach in terms of peak signal to noise ratio (PSNR) for image quality
assessment. The results showed that the proposed hybrid approach recorded higher PSNR
19.8 dB compared to the standalone adaptive median or wiener filters.
In addition, a parallel hybrid filter algorithm is also proposed for gray scale medical image
de-noising. The hybridization is between adaptive median and wiener filters. Parallelization
is applied on the adaptive median filter to overcome the latency of neighborhood operation,
parfor implicit parallelism powered by MATALAB 2013a is used. The algorithm is tested on
an image of 2.5 MB size, which is divided into 2, 4 and 8 partitions; a comparison between
the proposed algorithm and sequential one is given, in terms of time. Thus, each case has the
best time when assigned to number of threads equal to the number of its partitions. Moreover,
Speed up and efficiency are calculated for the algorithm and they record a measured
enhancement.