Reduction of impulse noise on grey scale image
Loading...
Date
2019-06
Authors
Aina Qistina Binti Md.Taha
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Digital images can easily be infected with noise. One of it is impulse noise (i.e.
salt-and-pepper noise). To deal with this type of noise is by using median filter. Median
filter is a type of non-linear filter and the median filter framework has been widely
extended with various kinds of methods. Standard median filter includes noisy pixels
in calculating the median value for the restoration process. Due to this can cause
inaccurate results because noisy pixel values may also be selected during restoration
process. Another extension of median filter is the recursive median filter. This type of
filter performs its calculation not only with the initial input pixel values but also
includes the output pixel values from previous filtering process. In this project, an
improvement to recursive median filter is proposed by adapting it with switching and
adaptive median filter. This proposed method is called as Recursive Switching
Adaptive Median Filter. Since the switching framework is included in the proposed
method, this method also consists of two stages, which are noise detection and noise
restoration stage. In the restoration stage, an adaptive method is used for restoration.
The size of the filter keeps expanding until at least 8 or more noise-free pixel
candidates are defined by the local window. The experimental results show good
performance in terms of Mean Square Error and Structural Similarity Index Measure.
However, the proposed method does not perform well at a high level of corruption.
Results shows that the proposed method shows good performance only in filtering
noised images with 0% up to 75% noise density. The performance of the proposed
filter decreases as the noise density in an image exceeds 80%.