ADAPTIVE APPROXIMATED MEDIAN FILTERING ALGORITHM FOR IMPULSE NOISE REDUCTION

Purchase PDF

Published: 2016-06-08

Page: 134-144


OBED APPIAH *

Department of Computer Science, Kwame Nkrumah University of Science and Technology, Ghana.

MICHAEL ASANTE

Department of Computer Science, Kwame Nkrumah University of Science and Technology, Ghana.

J. B. HAYFRON-ACQUAH

Department of Computer Science, Kwame Nkrumah University of Science and Technology, Ghana.

*Author to whom correspondence should be addressed.


Abstract

Most of the median based filtering algorithms have high computation time or cost due to the sorting subtask needed to select the median values. Given a relatively bigger window size such as 5 x 5 or 7 x7, the approach used in determining the median value to be used to replace the centre pixel can highly influence the total computational time required by the filter. The subtask required for median selection takes quite a significant computational time, coupled with the size of the image sometimes makes the algorithms slower hence, the implementation for real time applications becomes difficult. The need to have effective and efficient algorithms for filtering noise is always desirous since they will ensure effective implementation of various image processing tasks such as object identification, object tracking, and  others in the long round.  This paper proposes a new median based algorithm for the reduction of impulse noise in corrupted 2D greyscale images with relatively better computational time as well as quality results. The algorithm proposed is based on the concept of approximated median value determination which always takes lesser time in determining the median value as compared to the determination of the actual median. The median of medians method is used to select a median value, which then replaces the central pixel of the sliding window. The approximated median make it possible to use O(4/3n) to select a median values instead of O(n2) or O(n + k) as proposed by some median based filtering algorithms.  Seven different arrangements or partitions were used for the experiment, and their performance were very close, but the Greater Shape (> Shape) partition gave the best results.

Keywords: Median filtering, adaptive median filter, approximated median filtering, filtering, impulse noise


How to Cite

APPIAH, O., ASANTE, M., & HAYFRON-ACQUAH, J. B. (2016). ADAPTIVE APPROXIMATED MEDIAN FILTERING ALGORITHM FOR IMPULSE NOISE REDUCTION. Asian Journal of Mathematics and Computer Research, 12(2), 134–144. Retrieved from https://ikprress.org/index.php/AJOMCOR/article/view/592

Downloads

Download data is not yet available.